JP7545351B2 - Facility value estimation device, public facility impact assessment system, and public facility impact assessment method - Google Patents

Description

The present invention relates to a facility value estimation device, a public facility impact assessment system, and a public facility impact assessment method.

There is a system (public facility impact assessment system) that quantifies the impact of social infrastructure (such as water pipelines) on society by evaluating the value of various facilities in the local community from a specific perspective. In social infrastructure maintenance work using this public facility impact assessment system, when multiple parts of social infrastructure are damaged, it is necessary to make decisions about which parts should be repaired first.

These decisions are made based on the value of surrounding facilities that are affected by social infrastructure. However, quantifying the value of a facility is difficult. Therefore, Patent Document 1 describes a technology that quantifies the value of a facility by manually defining the value for each type of facility.

US Patent Application Publication No. 2020/0111039

The invention of Patent Document 1 attempts to solve the problem of quantifying the impact of infrastructure on society by having users manually quantify the value of each facility type, and then quantifying the impact on society by combining the quantified facility value, the facility's water usage, and repair time. In this regard, Patent Document 1 has users input facility value data for each type of facility, so it is thought that by inputting this data for each individual facility, it is possible to improve the accuracy of quantifying the facility value of each facility. However, it is thought that quantifying the value of each of the many facilities that exist in a city requires a lot of effort from the person inputting the data (user), and is not very effective.

For example, to efficiently quantify the value of facilities, it is possible to perform machine learning using past city data and facility evaluation values. However, the data required for machine learning is often insufficient because some information, such as information on individual facilities (e.g., the number of hospital beds), is not publicly available or is not organized. In addition, the teacher labels for facility value required for machine learning must be created manually, but judging the value of a facility is a highly specialized task, and it is difficult to verify their validity. Therefore, when using machine learning, it is also necessary to take into account the variability in teacher labels.

The present invention has been developed in consideration of these circumstances, and its purpose is to provide a facility value estimation device, a public facility impact assessment system, and a public facility impact assessment method that are capable of stably calculating highly accurate facility values.

One aspect of the present invention for solving the above-mentioned problems includes an estimation unit that has a processor and a memory, and reads attribute information of each facility belonging to the designated area and attribute information of the designated area from the memory that stores attribute information of a designated area, attribute information of each facility belonging to the designated area, and people flow data showing time-series movement of people in the designated area, and inputs attribute information of each facility belonging to the designated area and attribute information of the designated area into a first trained model that uses the attribute information of the facility and the attribute information of the designated area as input values and an evaluation value of the facility as an output value, thereby calculating an evaluation value of each facility belonging to the designated area and a calculation accuracy that is the reciprocal of the variance of the evaluation value, and a calculation unit that calculates the evaluation value of each facility belonging to the designated area by determining whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold. The facility value estimation device includes: a re-estimation unit that determines whether there is a low-accuracy facility, which is a facility whose accuracy is below a predetermined threshold, and if it is determined that there is a low-accuracy facility in the designated area, recalculates the evaluation value of the low-accuracy facility by inputting information on human connections between facilities in the designated area based on people flow data in the designated area read from the memory into a second trained model that uses information on human connections between facilities in the designated area as an input value and an evaluation value of the facility in the designated area as an output value; and a display unit that displays the calculated or recalculated facility evaluation value information, wherein the second trained model is created by a learning process that uses information on human connections between facilities in areas other than the designated area as teacher data for input values and the evaluation value of the facility in the other area as teacher data for output values .

Further, one of the present inventions for solving the above-mentioned problems includes an estimation unit having a processor and a memory , the estimation unit reading attribute information of each facility belonging to the designated area and attribute information of the designated area from the memory storing attribute information of a designated area, attribute information of each facility belonging to the designated area, and people flow data showing time-series movement of people in the designated area, and inputting the attribute information of each facility belonging to the designated area and the attribute information of the designated area into a first trained model having the attribute information of the facility and the attribute information of the designated area as input values and an evaluation value of the facility as an output value, thereby calculating an evaluation value of each facility belonging to the designated area and a calculation accuracy which is the reciprocal of the variance of the evaluation value, and a determination unit determining whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, thereby determining whether the designated area includes a low-accuracy facility which is a facility whose calculation accuracy of the evaluation value is less than a predetermined threshold, and when it is determined that the designated area includes a low-accuracy facility, the estimation unit determining whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, In this case, the public facility impact assessment system includes: a re-estimation unit that recalculates the assessment value of the low-accuracy facility by inputting information on human connections between facilities in the designated area based on people flow data in the designated area read from the memory into a second trained model that uses information on human connections between facilities in a specified area as an input value and an assessment value of the facility in the specified area as an output value; an evaluation unit that calculates an infrastructure impact degree that represents the magnitude of the impact of the public facility on each facility based on information on the layout of the public facilities, the calculated or recalculated facility value of each facility, and information on the location of each facility; and a display unit that displays the calculated or recalculated facility assessment value information and the calculated infrastructure impact degree , wherein the second trained model is created by a learning process that uses information on human connections between facilities in areas other than the designated area as teacher data for input values and the assessment value of the facility in the other areas as teacher data for output values .

Further, one aspect of the present invention for solving the above problem is an information processing device that reads attribute information of each facility belonging to the designated area and attribute information of the designated area from a memory that stores attribute information of a designated area, attribute information of each facility belonging to the designated area, and people flow data showing the time-series movement of people in the designated area, and inputs the attribute information of each facility belonging to the designated area and the attribute information of the designated area into a first trained model that uses the attribute information of the facility and the attribute information of the area to which the facility belongs as input values and the evaluation value of the facility as an output value, thereby performing an estimation process of calculating an evaluation value of each facility belonging to the designated area and a calculation accuracy that is the reciprocal of the variance of the evaluation value, and determining whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, thereby determining whether the designated area includes a low-accuracy facility that is a facility whose calculation accuracy of the evaluation value is less than a predetermined threshold, and if it is determined that the low-accuracy facility is included in the designated area, The public facility impact assessment method executes the following steps: a re-estimation process for recalculating the evaluation value of the low-accuracy facility by inputting information on human connections between facilities in a specified area based on people flow data in the specified area read from the memory into a second trained model having information on human connections between facilities in the specified area as an input value and an evaluation value of the facility in the specified area as an output value; an evaluation process for calculating an infrastructure impact degree representing the magnitude of the impact of the public facility on each facility based on information on the layout of the public facilities, the calculated or recalculated facility value of each facility, and information on the location of each facility; and a display process for displaying information on the calculated or recalculated evaluation value of the facility and the calculated infrastructure impact degree , wherein the second trained model is created by a learning process having information on human connections between facilities in areas other than the designated area as training data for input values and the evaluation value of the facility in the other areas as training data for output values .

The present invention makes it possible to stably calculate facility value with high accuracy.

Issues, configurations, and advantages other than those described above will become clear from the description of the embodiments below.

1 is a diagram showing an example of the configuration of a public facility impact assessment system according to an embodiment of the present invention; FIG. 2 is a diagram illustrating an example of functions provided in a facility value estimation device and a data management device. FIG. 2 is a diagram illustrating an example of city data. FIG. 11 is a diagram illustrating an example of people flow data. FIG. 2 is a diagram illustrating an example of infrastructure data. FIG. 2 is a diagram illustrating an example of hardware included in each information processing device. FIG. 11 is a sequence diagram illustrating an outline of a public facility impact assessment process. FIG. 11 is a sequence diagram illustrating an overview of a learning process for creating a first analytical model and a second analytical model. 13 is a flowchart illustrating an example of a facility value estimation process. FIG. 11 is a diagram illustrating an example of a facility value table. 11 is a flowchart illustrating an example of a data input process. 13 is an example of a data entry screen displayed in the data entry process. 13 is a flowchart illustrating details of a re-estimation process. FIG. 11 is an image diagram illustrating details of the re-estimation process. FIG. 13 is a diagram illustrating an example of an inter-facility connection relationship table. FIG. 13 illustrates an example of a network structure table. FIG. 13 illustrates an example of a new network structure table. 11 is a flowchart illustrating an example of an infrastructure impact degree calculation process. 11 is a diagram illustrating an example of an infrastructure impact degree calculation process when the infrastructure is a water pipeline; FIG. FIG. 13 illustrates an example of an infrastructure impact table. FIG. 13 is a diagram showing an example of a model learning screen that is displayed when learning a first analytical model. FIG. 13 is a diagram showing an example of an estimation result display screen. FIG. 13 is a diagram showing an example of a re-estimation result display screen. FIG. 13 is a diagram showing an example of a final estimation result display screen. FIG. 13 is a diagram illustrating an example of an impact evaluation display screen. 11 is a flowchart illustrating an example of a disaster recovery support process. 11 is a flowchart illustrating an example of an annual maintenance plan support process. FIG. 13 is a diagram showing an example of integrated risk assessment information.

The following describes an embodiment of the present invention with reference to the drawings. Parts with the same reference numerals represent the same objects, and the basic configurations and operations are similar.

--System configuration--
1 is a diagram showing an example of the configuration of a public facility impact assessment system 1 according to this embodiment. The public facility impact assessment system 1 includes a facility value estimation system 30 having a facility value estimation device 10 and a data management device 20, and an external system 40 having a sensor state detection system 113, a damage risk prediction system 114, and a repair and restoration plan formulation system 115.

The facility value estimation device 10, data management device 20, sensor status detection system 113, damage risk prediction system 114, and repair and restoration plan formulation system 115 are connected via wired or wireless communication networks 5 and 6 such as a LAN (Local Area Network), a WAN (Wide Area Network), the Internet, or a dedicated line.

The public facility impact assessment system 1 is an information processing system that manages public facilities (public infrastructure facilities such as water, gas, electricity, etc.; hereafter simply referred to as infrastructure) installed in each area (city, etc.).

When managing infrastructure, the public facility impact assessment system 1 estimates the evaluation value (hereinafter referred to as facility value) of each facility in each area, including the area (hereinafter referred to as the designated area) specified by the user (system administrator, etc.), in relation to the infrastructure, for each area, using two analytical models described below.

Then, based on the estimated facility value, the public facility impact assessment system 1 calculates the infrastructure impact, which is information on the impact of the infrastructure on each facility within the specified area (for example, the impact of infrastructure maintenance on each facility).

Next, FIG. 2 is a diagram illustrating an example of the functions of the facility value estimation device 10 and the data management device 20.

The facility value estimation device 10 includes a display unit 101, an estimation unit 102 that estimates the facility value, a determination unit 103 that determines the calculation accuracy of the facility value estimated by the estimation unit 102, a re-estimation unit 104 that performs a re-estimation when the calculation accuracy of the facility value is low, a calculation unit 105 that performs a calculation process of the facility value based on the estimation unit 102 or the re-estimation unit 104, an evaluation unit 106 that calculates the infrastructure impact degree, and an application unit 117 that includes a public facility management unit 116 that manages public facilities.

The facility value estimation device 10 also includes a base unit 118 that includes a storage unit 107 that stores data and a setting unit 108 that sets various data, a communication unit 112 that communicates with an external system 40, and a communication unit 119 that communicates with the data management device 20.

The estimation unit 102 calculates the evaluation value (facility value) of each facility and also calculates the calculation accuracy of the evaluation value. In this embodiment, the estimation unit 102 calculates the evaluation value and the calculation accuracy by using the first trained model for each area.

That is, the estimation unit 102 inputs attribute information of a specified facility and attribute information of the area to which the facility belongs into a first trained model (hereinafter referred to as the first analytical model) that uses attribute information of the facility and the area to which the facility belongs as input values and an evaluation value of the facility as an output value, thereby calculating the evaluation value and the calculation accuracy of the evaluation value of the specified facility.

The re-estimation unit 104 determines whether there are any facilities (low-accuracy facilities) for which the calculation accuracy of the evaluation value calculated by the estimation unit 102 is less than a predetermined threshold, and if there are any low-accuracy facilities, identifies other groups (communities) of multiple facilities that are highly similar to the group (community) of multiple facilities including the low-accuracy facility in predetermined elements (described below), and recalculates the evaluation value of the low-accuracy facility based on the evaluation value of the facilities in the other identified groups.

For example, with regard to the similarity of the above-mentioned specified elements, the re-estimation unit 104 calculates the evaluation value of a low-accuracy facility based on the evaluation value of a facility in another group of multiple facilities that has a human connection that is highly similar to the human connection (such as the flow of people between facilities) between facilities in the group of multiple facilities that includes the low-accuracy facility.

In this embodiment, the re-estimation unit 104 recalculates the evaluation value and calculation accuracy in the specified area by using the second trained model for each area, including the specified area.

In other words, when there is a low-accuracy facility, the re-estimation unit 104 recalculates the evaluation value of the low-accuracy facility by inputting information on the human connections (such as people flow between facilities) between facilities in a specified area (area other than the specified area) into a second trained model (hereinafter referred to as the second analysis model) that uses information on the human connections between facilities in an area that includes the low-accuracy facility (i.e., the specified area) as an input value and the evaluation value of the facility in the specified area as an output value. Note that this human connection is the network that forms the basis of the above-mentioned community, as will be described later.

The re-estimation unit 104 also accepts input from the user for further corrections to the recalculated evaluation value (facility value) of the low-accuracy facility.

Next, the evaluation unit 106 performs calculations regarding the impact that the infrastructure has on each facility.

For example, the evaluation unit 106 calculates an infrastructure impact degree that indicates the magnitude of the impact of public facilities on facilities based on information on the location of public facilities, the facility value of each facility calculated by the estimation unit 102 and the re-estimation unit 104, and information on the location of each facility.

The public facilities management unit 116 performs calculations related to infrastructure management.

For example, the public facility management unit 116 obtains information on abnormal parts of public facilities from the sensor status detection system 113, and creates information (recovery plan information) indicating the need to repair the abnormal parts based on the impact on the infrastructure.

In addition, the public facility management unit 116 obtains information on the risk of damage to public facilities from the damage risk prediction system 114, and calculates an assessment value (integrated risk assessment value) indicating the degree of need to respond to the risk of damage based on the infrastructure impact degree.

Next, the display unit 101 displays various information on the screen, such as the facility value of each facility and its impact on infrastructure.

For example, the display unit 101 displays information on the evaluation values of each facility (facilities that are not low-accuracy facilities) whose evaluation value calculation accuracy is equal to or greater than a predetermined threshold, and information on the recalculated evaluation values of low-accuracy facilities. At this time, the display unit 101 displays information on facilities whose evaluation value calculation accuracy is equal to or less than the predetermined threshold in a first display format, and displays information on the other facilities in a second display format.

Next, the data management device 20 stores various data including city data 109, which is data on the attributes of each facility in each area, people flow data 110, which is data on the attributes of people in each area (data on the location of each person at each date and time), and infrastructure data 111, which is data on the attributes of infrastructure in each area.

(City Data)
FIG. 3 is a diagram showing an example of the city data 109. The city data 109 has data items of a facility ID 1091, a facility name 1092, a facility type 1093, a facility latitude 1094, a facility longitude 1095, a facility area 1096, a facility employee number 1097, a facility capacity 1098, a facility bed number 1099, a facility annual sales 10991, and a facility public land price 10992. The data items described here are examples, and some of these data items may be omitted or other items may be added. In addition, there may be missing values in the data of the city data 109 due to reasons such as not being disclosed to a specific organization. In addition, the types of facilities include, for example, hospitals, schools, childcare facilities, nursing care facilities, evacuation shelters, houses, office buildings, factories, and industrial facilities.

(People flow data)
4 is a diagram showing an example of people flow data 110. The people flow data 110 has the data items of a user ID 1101, date and time 1102, latitude 1103, and longitude 1104. This data is acquired from location information or the like emitted by a terminal such as a mobile phone. Note that the people flow data 110 may be information indicating the chronological movement of people. Note that the data items described here are merely examples, and some of these data items may be omitted, or other items may be added.

(Infrastructure Data)
5 is a diagram showing an example of the infrastructure data 111. In this embodiment, the infrastructure data 111 is data related to water pipelines in each area, but as will be described later, the present invention is also applicable to other infrastructures.

That is, the infrastructure data 111 in this embodiment has a water demand table 201, a pipeline network structure table 202, and a teacher label 203.

The water demand table 201 is data on the demand for water for each facility. The water demand table 201 has the following data items: the ID 2011 of the facility that uses water, the average water usage for that facility 2012, and the average water usage charge for that facility 2013.

The pipeline network structure table 202 is data on the location and layout of water pipelines. The pipeline network structure table 202 has the following data items: pipeline ID 2021, pipeline start node ID 2022 (one end of the pipeline), pipeline end node ID 2023 (the other end of the pipeline), pipeline latitude 2024, pipeline start node longitude 2025, pipeline end node latitude 2026, and pipeline end node longitude 2027. Here, a node corresponds to a connection between pipelines.

The teacher label 203 is a teacher label (correct answer data) of the facility value of each facility, which is used for creating (learning) each analysis model. In this embodiment, the facility value is the sum of the human value and the economic value of the facility.

That is, the teacher label 203 has the data items of the facility ID 2031, the facility's human value 204, and the facility's economic value 205. The teacher label 203 is set, for example, by an infrastructure service provider, etc.

Note that the data items of the infrastructure data 111 described above are merely examples, and some of these data items may be omitted or other items may be added. For example, the infrastructure data 111 may be data indicating the usage and structure of the infrastructure.

Next, as shown in FIG. 1, the sensor status detection system 113 identifies abnormal locations in the public facilities by acquiring information from sensors that have been installed in advance at various locations in the public facilities.

The damage risk prediction system 114 calculates the damage probability of public facilities corresponding to the time axis. The damage probability is an evaluation value that indicates the degree of need for repairs to abnormal parts of the public facilities.

The repair and restoration planning system 115 displays information on the screen regarding the repair or restoration of public facilities during normal times or when a disaster occurs.

6 is a diagram showing an example of hardware provided in each information processing device of the facility value estimation device 10, the data management device 20, the sensor state detection system 113, the damage risk prediction system 114, and the repair and restoration plan planning system 115. Each information processing device includes a processing device 11 such as a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM,
A main memory 12 such as a read only memory (Read Only Memory), a hard disk drive (HDD), a solid state drive (SSD),
The computer includes an auxiliary storage device 13 such as a centralized state drive (CSD), an input device 14 such as a keyboard, a mouse, a touch panel, an output device 15 such as a monitor (display), and a communication device 16 for communicating with other devices.

The processing performed by each information processing device is realized by the processing device 11 executing a program stored in the main memory device 12. These programs are stored in, for example, a secondary memory device, a non-volatile semiconductor memory, a hard disk drive, an SSD, or other storage device, or a non-transitory data storage medium that can be read by each node, such as an IC card, an SD card, or a DVD.

Next, we will explain the processing performed by the public facility impact assessment system 1.

--Processing Overview--
First, an overview of the processing performed by the public facility impact assessment system 1 will be described.

Figure 7 is a sequence diagram outlining the public facility impact assessment process, which estimates the facility value of each facility and assesses the impact on public facilities. The public facility impact assessment process is executed, for example, after the execution of the learning process described below, or together with the learning process, when a predetermined input is made to the facility value estimation device 10 by the user.

First, the estimation unit 102 of the facility value estimation device 10 acquires data from the data management device 20 (s1), and calculates the facility value and the facility value calculation accuracy of each facility based on the acquired data (s2). At this time, the estimation unit 102 calculates the facility value and the facility value calculation accuracy using the first analysis model.

Based on the information of the processing result of the estimation unit 102, the re-estimation unit 104 recalculates the facility value of the facility with low calculation accuracy based on the community extraction method described below (s4). At this time, the re-estimation unit 104 recalculates the facility value by inputting the data of the community into the second analysis model.

The evaluation unit 106 calculates the infrastructure impact based on the facility value calculation results (and recalculation results) of each facility by the estimation unit 103 and the re-estimation unit 104, and information on the layout and configuration of the public facilities (network between facilities) (s6), and stores the calculation results of the infrastructure impact in a specified table (s7). This completes the public facility impact evaluation process.

Next, FIG. 8 is a sequence diagram outlining the learning process for creating the first and second analytical models.

As shown in the figure, the estimation unit 102 of the facility value estimation device 10 performs predetermined machine learning based on the data acquired from the data management device 20 to create a first analytical model (s11). In addition, the re-estimation unit 104 of the facility value estimation device 10 creates an inter-facility network based on the data acquired from the data management device 20, reconstructs the created inter-facility network into communities (groups) using a community extraction method described below, and performs predetermined machine learning based on the data of the reconstructed communities to create a second analytical model (s12). The learning process then ends.

Next, we will explain the value estimation process performed by the facility value estimation device 10 to estimate the facility value of each facility.

- Facility value estimation processing -
FIG. 9 is a flowchart illustrating an example of the facility value estimation process.

First, the estimation unit 102 of the facility value estimation device 10 executes a data input process to set each piece of data in the data management device 20 (s301). The estimation unit 102 also accepts the designation of the area (designated area) for which the user wishes to calculate the facility value. The details of the data input process will be described later.

The estimation unit 102 creates a first analytical model for evaluating the facility value, and calculates an evaluation value of the facility value of each facility in the designated area using the created first analytical model (s302). In this embodiment, the estimation unit 102 creates a first analytical model related to human value (human first analytical model) and a first analytical model related to economic value (economic first analytical model) for each area.

For example, first, the estimation unit 102 performs a learning process (e.g., using past city data 109 and infrastructure data 111) for each area, using each data of the city data 109 as teacher data for input values and each data of the human value 204 of the teacher label 203 of the infrastructure data 111 as teacher data for output values, thereby creating a first human analysis model. Note that this analysis model is constructed, for example, as a neural network consisting of an input layer, an intermediate layer, and an output layer. Note that the output layer outputs a probability distribution of output values (probability distribution having a mean and variance). Such a neural network is created, for example, by Bayesian linear regression or Gaussian process regression.

Similarly, the estimation unit 102 creates a first economic analysis model for each area by performing a learning process (e.g., using past city data 109 and infrastructure data 111) in which each data of the city data 109 is used as teacher data for input values and each data of the economic value 205 of the teacher label 203 of the infrastructure data 111 is used as teacher data for output values. Note that these first analysis models may be created before executing the facility value estimation process (e.g., in the learning process).

The estimation unit 102 then inputs the current city data 109 for each facility in the specified area into the first human analysis model and the first economic analysis model created in this way, and outputs the probability distribution of the human value evaluation value (human evaluation value) and the probability distribution of the economic value evaluation value (economic evaluation value) for each corresponding facility.

Then, the determination unit 103 determines the calculation accuracy of each facility value for each first analysis model based on the evaluation value output in s302. That is, the determination unit 103 determines whether the calculation accuracy of each facility value is equal to or greater than a predetermined threshold value (s303).

For example, the determination unit 103 calculates the inverse of the variance of the probability distribution of the human evaluation value output in s302 and the inverse of the variance of the probability distribution of the economic evaluation value, thereby calculating the calculation accuracy of the facility value using the first human analysis model and the first economic analysis model. The determination unit 103 determines whether the calculation accuracy of each facility value using these first analysis models is equal to or greater than a predetermined threshold value.

For each first analysis model, if the calculation accuracy of the facility value of each facility is equal to or greater than a predetermined threshold (s303: high accuracy), the determination unit 103 executes processing of s304 for the facility value of that facility, and if the calculation accuracy of the facility value of each facility is less than the predetermined threshold (s303: low accuracy), the determination unit 103 executes re-estimation processing s305 for the facility value of that facility (described below). After that, processing of s304 is performed.

In s304, the determination unit 103 sets each evaluation value output in s302 in the facility value table 702.

(Facility value table)
10 is a diagram showing an example of a facility value table 702. The facility value table 702 has the following data items: a facility ID 7021, a human value average 7022 of the facility (average value of human evaluation values), a calculation accuracy 7023 of the human evaluation value of the facility, an economic value average 7024 of the facility (average value of economic evaluation values), a calculation accuracy 7025 of the economic evaluation value of the facility, a facility value average 7026 of the facility (details will be described next), and a facility value accuracy 7027 of the facility (details will be described next). The facility value average 7026 and the facility value accuracy 7027 are set by process s306, which will be described next.

Next, as shown in s306 of FIG. 9, the calculation unit 105 calculates the final facility value of each facility and sets it in the facility value table 702. That is, this facility value is calculated as a linear sum of the average human evaluation value and the average economic evaluation value. For example, the calculation unit 105 calculates the facility value of each facility using the following formula, using weighting coefficients a and b that indicate the relative importance of human elements and economic elements.

Facility value = (a x average human evaluation value) + (b x average economic evaluation value)
(a is the weighting coefficient of the human factor, b is the weighting coefficient of the economic factor, a + b = 1)

The facility value accuracy is calculated as the reciprocal of the linear sum of the reciprocal of the calculation accuracy of the average human evaluation value and the reciprocal of the calculation accuracy of the average economic evaluation value. This is based on the fact that the value of accuracy is the reciprocal of variance, and the variance of the sum of two independent probability distributions is equal to the sum of the variance of each probability distribution.

In this way, by considering the weighting coefficients a and b for each evaluation value, it becomes possible to evaluate the value of facilities according to the economic and social conditions of the city. For example, during normal times, the weighting coefficients are set as (a, b) = (0.2, 0, 8) to prioritize infrastructure connected to facilities that incur large economic losses, and during disasters, the weighting coefficients are set as (a, b) = (0.8, 0.2) to prioritize infrastructure connected to facilities that are critical to human life.

Next, the evaluation unit 106 executes an infrastructure impact calculation process (s307) to calculate the infrastructure impact based on the facility value of each facility calculated in s306. The display unit 101 displays information on the calculated infrastructure impact on the screen (s308).

Here, we will explain the details of the data input process s301, the re-estimation process s305, and the infrastructure impact calculation process s307.

- Data entry processing -
Fig. 11 is a flow chart for explaining an example of the data input process, and Fig. 12 is an example of a data input screen 1101 displayed in the data input process.

As shown in FIG. 11, the setting unit 108 receives an area selection from the user via the data input screen 1101 (s1107). The setting unit 108 acquires city data 109 and people flow data 110 related to the selected area from the data management device 20 (s1108). The setting unit 108 also accepts input of infrastructure data 111 from the user via the data input screen 1101. The storage unit 107 sets the acquired or input data to the city data 109, people flow data 110, and infrastructure data 111 (s1109).

As shown in FIG. 12, the data input screen 1101 includes an area selection field 1103 for accepting an area selection from the user, a pipeline structure data selection field 1104 for reading the pipeline network structure table 202 of the infrastructure data 111, a water usage charge data selection field 1105 for reading the water demand table 201 of the infrastructure data 111, and a create new area button 1106 for completing the setting of data related to the selected area. The data input screen 1101 may also include fields for editing the city data 109 and people flow data 110.

- Re-estimation process -
FIG. 13 is a flowchart for explaining the details of the re-estimation process S305.
FIG. 13 is an image diagram illustrating the details of the re-estimation process S305.

As shown in these figures, the re-estimation unit 104 acquires the people flow data 110 and the facility value table 702 created in s304 (s401). The re-estimation unit 104 then creates the facility connection relationship table 601, which is data that integrates the people flow data 110 and the facility value table 702 (s402).

Furthermore, the re-estimation unit 104 converts the facility-to-facility connection relationship table 601 into a network structure table 602 that includes elements of nodes 504 indicating each facility, edges 503 indicating the people flow relationships between each facility, and weights 505 indicating the strength of the people flow relationships (strength of human ties) (s403).

As shown in FIG. 14, the edge weight 505 is set to a value that represents the strength of the connection between nodes 504. For example, if the proportion of people living in "House C" who go to "Hospital A" is small but the proportion of people who go to "Hospital B" is large, the connection relationship between nodes 504 is stronger between "House C" and "Hospital B" than between "House C" and "Hospital A", so the value of the weight 505 of the edge 503 between "House C" and "Hospital B" (5.2) is greater than the weight value between "House C" and "Hospital A" (0.9).

Here, we will explain the details of the facility connection relationship table 601 and the network structure table 602.

(Inter-facility connection relationship table)
15 is a diagram showing an example of an inter-facility connection relationship table 601. The inter-facility connection relationship table 601 stores data on people flow between facilities. The inter-facility connection relationship table 601 has data items of an ID 6011 of a facility that is the start point between the facilities, an ID 6012 of a facility that is the end point between the facilities, and the number of people moving between these facilities per unit time 6013. Note that the data items described here are merely examples, and some of these data items may be omitted or other items may be added.

(Network Structure Table 602)
FIG. 16 is a diagram showing an example of the network structure table 602. The network structure table 602 stores the facility value of each facility along with data on people flow between facilities. That is, the network structure table 602 is information showing the dependency (network) between facilities based on human connections. Specifically, the network structure table 602 has data items of an edge ID 6021 which is an ID of the relationship (edge) between facilities, a node ID 6022 (similar to a facility ID) at the start point of the edge, a node ID 6023 (similar to a facility ID) at the end point of the edge, a weight 6024 of the edge, an average facility value 6025 of the start node, and an average facility value 6026 of the end node. The weight 505 is calculated based on the people flow between each facility (node) in the facility connection relationship table 601.

Note that the facility value average 6025 of the starting node and the facility value average 6026 of the ending node may or may not be set to a value. That is, for facilities (nodes) that are determined to have high estimation accuracy in the facility value estimation process (s303: high accuracy), facility value information for the facility is set (data is obtained from the facility value table 702). On the other hand, for facilities that are determined to have low estimation accuracy in the facility value estimation process (s303: low accuracy), no value is set (it is blank), and a value is set by the re-estimation process described below.

13 and 14, the re-estimation unit 104 reconstructs the facilities (nodes) in each area, including the designated area, into a plurality of groups (communities 507) based on the network structure table 602, thereby creating a new network structure table 603. For example, the re-estimation unit 104 performs the reconstruction based on a community extraction method. Note that, as will be described later, the creation of the new network structure table 603 for areas excluding the designated area and the training of the second analysis model based on this may be performed in advance.

(New network structure table)
17 is a diagram showing an example of a new network structure table. The new network structure table 603 is a table in which community information has been added to the network structure table 602. The new network structure table 603 has the following data items: an edge ID 6031, an edge start node ID 6032, an edge end node ID 6033, an edge weight 6034, an average facility value 6035 of the start node, an average facility value 6036 of the end node, and an ID 6037 of the community to which the edge belongs.

The above-mentioned community extraction method is a method for extracting densely connected subsets from a network structure. Examples of community extraction methods include label propagation, modularity optimization, and community extraction methods based on edge betweenness centrality. The community extraction methods listed here are only examples, and the method is not particularly limited as long as it is possible to extract a subset from a network structure. In this embodiment, a community extraction method based on edge betweenness centrality is used. That is, the re-estimation unit 104 calculates edge betweenness centrality using the network structure data 602, and creates a new network structure table 603 by repeatedly removing edges with the maximum edge betweenness centrality to extract densely connected communities (groups of multiple facilities).

There are two reasons for adopting the community extraction method. The first is to make it easier to extract the relationships between facilities within a community by dividing the population into communities with dense connections. The re-estimation process s504 estimates the unknown facility value using a second analysis model based on the feature of the relationship between facilities. If the re-estimation process s504 is executed without community extraction, many loosely connected relationships will be included, making it difficult to extract the feature that expresses the relationship between facilities. Therefore, this problem is addressed by dividing the population into communities with dense connections using the community extraction method and using them in the facility value estimation process (s405) using the second analysis model described later. The second is that by dividing the population into communities, the same second analysis model can be used for different areas. For example, the number, size, or relationship between facilities in the area is different between "area A" used for machine learning of the analysis model and "area B" where the facility value is to be estimated using that analysis model, so it is generally difficult to estimate the facility value using information for the entire area. Here, it can be assumed that in each densely connected community, the number or size of facilities will differ, but the relationships between facility values will be roughly the same. For example, a hospital that has strong connections with many nearby residential facilities is important, and such relationships will be similar across different communities or areas. According to this assumption, in each densely connected community, the relationships between facilities will be roughly the same. As a result, even if the areas are different, it will be possible to estimate facility values using the same model on a community basis.

Next, as shown in FIG. 13, the re-estimation unit 104 estimates the facility value of facilities with low estimation accuracy (low-accuracy facilities) based on the new network structure table 603 created in s404 (s405).

Specifically, for example, the re-estimation unit 104 creates a second analysis model by performing a learning process in which the type of facility corresponding to the start node ID 6032 of each record in the new network structure table 603, the type of facility corresponding to the end node ID 6033, and the edge weight 6034 are used as teacher data for input values, and the average value 6036 of the facility value of the end node of each record in the new network structure table 603 is used as teacher data for output values for areas other than the designated area. Note that this second analysis model is constructed as a graph Bayesian neural network consisting of, for example, an input layer, an intermediate layer, and an output layer. Note that the output layer outputs a probability distribution of output values (probability distribution having a mean and a variance). Note that the second analysis model and the new network structure table 603, which is the premise of the second analysis model, may be created in advance (learning process).

Then, the re-estimation unit 104 calculates the average facility value and facility value accuracy of low-accuracy facilities by inputting the data of the new network structure table 603 for the designated area (excluding the data of records related to low-accuracy facilities) created in s404 into the second analysis model for areas other than the designated area.

Note that if there are multiple areas other than the specified area, the re-estimation unit 104 uses, for example, the output result of the second analysis model with the highest calculation accuracy.

The re-estimation unit 104 updates the information of the low-accuracy facilities estimated in s405 with the estimation results from the second analysis model (s406).

For example, the re-estimation unit 104 sets the recalculated facility value average to the facility value average 6025 of the start node and the facility value average 6026 (which is blank) of the end node of the record related to the low-accuracy facility in the network structure table 602. The re-estimation unit 104 also sets the re-estimated facility value average and facility value accuracy to the facility value average 7026 and facility value accuracy 7027 of the record related to the low-accuracy facility in the facility value table 702. This completes the re-estimation process.

The re-estimation unit 104 also accepts input from the user to correct information such as facility value for the estimation accuracy of the re-estimated facility value, and updates the facility value table 702 with the input data. In this case, since the second analysis model outputs information on the estimation accuracy of the facility value, it is clear to the user what needs to be corrected. This allows the user to set a more appropriate facility value with less effort (amount of correction).

- Infrastructure impact calculation process -
Next, Fig. 18 is a flowchart illustrating an example of the infrastructure impact calculation process S307. Fig. 19 is a diagram illustrating an example of the infrastructure impact calculation process when the infrastructure is a water pipeline.

First, the evaluation unit 106 acquires the coordinates of each facility from the city data 109, and calculates the representative coordinates 905 (representative point) of each pipeline 903 based on the pipeline network structure table 202 (s801). In this embodiment, the representative coordinates 905 are the coordinates of the midpoint between the start point and end point of the pipeline, but are not limited to this.

The evaluation unit 106 calculates the surrounding facility impact degree for each pipeline based on the coordinate data acquired in s801 (s802). The evaluation unit 106 sets the calculated surrounding facility impact degree in the infrastructure impact degree table 907 (s806).

The influence of surrounding facilities is the sum of the average facility values of each facility 904 close to the representative coordinates of the pipeline (e.g., facilities within a specified distance from the representative coordinates of the pipeline).

For example, in the case where there is pipeline A, pipeline B, pipeline C, house A, house B, house C, and hospital A as shown in Figure 19, house A is the calculation factor for the influence of pipeline A, house B is the calculation factor for pipeline B, and hospital A and house C are the calculation factors for the influence of pipeline C on surrounding facilities.

As shown in FIG. 18, the evaluation unit 106 calculates the connection impact based on the surrounding facility impact calculated in s802 (s803, s804). For example, the evaluation unit 106 calculates the product of a matrix indicating the surrounding facility impact of each pipeline and a reachability matrix indicating the connection impact. The evaluation unit 106 sets the calculated connection impact in the infrastructure impact table 907 (s806).

The connection influence degree is a parameter indicating the magnitude of the influence that a certain pipeline has on other pipelines. In this embodiment, the connection influence degree is a parameter indicating the influence that a certain pipeline has on its downstream pipeline. For example, as shown in Fig. 19, pipeline A can reach pipelines B and C, so the corresponding matrix element of the reachability matrix 902 is 1, and the reverse is not true due to the relationship in the flow direction, so the corresponding matrix element is 0.

Furthermore, the evaluation unit 106 calculates the infrastructure impact by adding the surrounding facility impact calculated in s802 and the connection impact calculated in s804 (s805). The evaluation unit 106 sets the calculated infrastructure impact in the infrastructure impact table 907 (s806).

(Infrastructure Impact Table)
20 is a diagram showing an example of the infrastructure impact table 907. The infrastructure impact table 907 has the following data items: a pipeline ID 9071, a pipeline surrounding facility impact 9072, a pipeline connection impact 9073, and a pipeline impact 9074.

The following describes the screens displayed by the facility value assessment device 10, etc., during the public facility impact assessment process.

(Model learning screen)
21 is a diagram showing an example of a model learning screen 1000 displayed in a learning process related to a first analytical model. The model learning screen 1000 includes a data input screen 1001 that accepts input of teacher data required for learning from a user, a learning button 1005 that accepts execution of learning from the user, a verification button 1006 that accepts execution of the first analytical model after learning from the user, and a verification result screen 1007 and a data list 1008 that accept display of the execution result of the first analytical model from the user.

The data input screen 1001 includes an area selection field 1002 that accepts the selection of a specified area and reads the city data 109 and people flow data 110 related to the specified area, a water data selection field 1003 that accepts the selection of the water demand table 201 of the infrastructure data 111, and a teacher label selection field 1004 that accepts the selection of the teacher label 203.

When data has been entered on the data input screen 1001, and the learn button 1005 and verify button 1006 are selected in that order, the verification results are output to the verification result screen 1007 and data list 1008.

The verification result screen 1007 displays information on the verification results regarding the first analytical model. For example, the mean squared sum error or average accuracy of the facility value of all facilities in the area according to the first analytical model is output. Note that the information displayed on the verification result screen 1007 is not limited to these, and is not particularly limited as long as it is information regarding the verification results of the first analytical model.

Furthermore, a data list of facilities in the area with low estimation accuracy (low-accuracy facilities) is displayed in the data list 1008. Since a facility with low estimation accuracy is likely to be a facility with insufficient data, the user can check the facility displayed in the data list 1008 and add information about the facility to the city data 109, thereby additionally learning and updating the first analysis model.

Next, we will explain the screens that are displayed during the facility value estimation process.

- Estimation result display screen -
22 is a diagram showing an example of an estimation result display screen 1201. The estimation result display screen 1201 is displayed, for example, during the processing of S304. The estimation result display screen 1201 displays a map 1203 of the specified area and , and a facility value estimation result table 1204. Facilities 1205, water pipelines 1206, and facility details information 1207 are displayed on the map 1203. The facilities 1205 are displayed by icons according to their types. The icon is colored according to the magnitude of the facility value (reference numeral 1208). If the estimated facility value of the facility 1205 is low in accuracy, a message indicating the low accuracy is displayed near the icon. For example, the lower the accuracy, the more marks (<*>) are displayed.

The facility value estimation result table 1204 displays the following items: facility ID, facility value, facility value accuracy, and judgment result. A predetermined color 1211 is applied to the display column of a facility that has been determined to have a low estimation accuracy of the facility value by the facility value estimation process (low-accuracy facility), and a display corresponding to facility 1205 on map 1203 is also performed. On the other hand, no coloring or map 1203 is displayed in the display column of facilities other than low-accuracy facilities. In this way, by varying the display format, information can be presented to the user in an easy-to-understand manner. Note that instead of coloring, any display form that visually differentiates the information, such as a shading display, can be used.

In addition, on the map 1203, detailed facility information 1207 for a facility 1205 is displayed when the icon for that facility 1205 is selected.

In addition, when any icon on the map 1203 or any facility on the facility value estimation result table 1204 is selected, the user can modify the facility value of the corresponding facility.

In addition, when the revision proposal display button 1209 is selected, a re-estimation (re-estimation process S305) of the facility value is performed for low-precision facilities (facilities with coloring 1211), and then the re-estimation result display screen 1202, which will be described next, is displayed.

- Estimation result display screen -
23 is a diagram showing an example of a re-estimated result display screen 1202. The re-estimated result display screen 1202 displays a map 1203 of a specified area, similar to the estimation result display screen 1201, and a facility value estimation result table 1204. At this time, in the facility value estimation result table 1204, the facility value and the estimation accuracy of the low-accuracy facility are corrected by re-estimation (reference numerals 1212 and 1213). However, even after re-estimation, the estimated accuracy of the facility value is still low. For facilities 1213 that are determined to be above the threshold, the user can use the map 1203 to correct and input the facility value.

In this way, the user only needs to check the facility value of facilities with low estimated facility value accuracy, reducing the amount of work required. After that, for example, if the user edits all the facilities whose facility value accuracy has been determined to be low, the final estimated result display screen 1301, which will be described next, is displayed.

-Final estimation result display screen-
24 is a diagram showing an example of a final estimation result display screen 1301. The final estimation result display screen 1301 displays a map 1304 of the designated area, a facility value estimation result table 1305, and an impact An impact evaluation button 1303 is displayed. When the impact evaluation button 1303 is selected, the infrastructure impact calculation process 307 is executed, and an impact evaluation display screen 1302 is displayed.

- Impact assessment display screen -
25 is a diagram showing an example of an impact evaluation display screen 1302. The impact evaluation display screen 1302 displays a map 1304 of a specified area similar to the final estimation result display screen 1301, and a pipeline impact evaluation result 1305. The pipeline impact evaluation result 1305 displays the pipeline ID of each pipeline, the pipeline impact calculated by the infrastructure impact calculation process, and the impact level obtained by discretizing the pipeline impact, in descending order of impact level. The impact level is a value discretized according to the pipeline impact, and is discretized into three "impact levels" in the example shown in the figure. Note that the discretization index based on the pipeline impact is not limited to this "impact level".

The map 1304 of the specified area displays pipelines 1306 colored according to the impact level, and detailed pipeline information 1307. Using the map 1304 of the specified area and the pipeline impact assessment results 1305, the user can determine which infrastructure has a large impact on the facility and use this information for subsequent infrastructure maintenance or restoration plans, etc.

Next, we will explain the processes related to the management of public facilities (disaster recovery support process and annual maintenance plan support process) that utilize the facility value estimation process.

- Disaster recovery support processing -
FIG. 26 is a flowchart illustrating an example of a disaster recovery support process, which is a process for performing disaster recovery by utilizing the facility value estimation process.

First, the estimation unit 102 of the facility value estimation device 10 acquires information on the pipeline in which an abnormality has occurred from the sensor state detection system 113. Specifically, the sensor state detection system 113 identifies the abnormal location in the water pipeline by acquiring information from vibration sensors that have been attached in advance to various locations in the water pipeline. The estimation unit 102 receives information on the abnormal location identified by the sensor state detection system 113.

Then, the facility value estimation device 10 executes a facility value estimation process (s1402). For example, the facility value estimation device 10 acquires data from the data management device 20 and estimates the facility value of each facility (estimation process). For facilities whose facility value is determined to be low in accuracy, the facility value estimation device 10 re-estimates the facility value (re-estimation process). The facility value estimation device 10 creates a facility value table 702 based on the results of the estimation process and the re-estimation process. At this time, the facility value estimation device 10 calculates the facility value of each facility based on weights a and b, which are set to appropriate values according to the disaster situation. The facility value estimation device 10 calculates the infrastructure impact of the abnormal pipeline based on the facility value of each facility and the infrastructure impact table 907.

The facility value estimation device 10 displays the infrastructure impact degree calculated in s1402. The facility value estimation device 10 also transmits information on the calculated infrastructure impact degree to the repair and restoration plan formulation system 115 (s1403). The repair and restoration plan formulation system 115 creates information on a restoration plan for the infrastructure based on this information and displays this on the screen. This enables the user to create a restoration plan for the water pipeline so as to minimize the impact on the water pipeline system.

- Annual maintenance planning support processing -
FIG. 27 is a flowchart illustrating an example of an annual maintenance plan support process, which is a process for making an annual maintenance plan under normal circumstances by utilizing the facility value estimation process.

First, the estimation unit 102 of the facility value estimation device 10 acquires information on the probability of breakage of each pipeline from the breakage risk prediction system 114. Specifically, the breakage risk prediction system 114 calculates the probability of breakage of each water pipeline based on information from sensors, etc. acquired from the sensor status detection system 113, etc. The estimation unit 102 receives the information on the probability of breakage acquired by the breakage risk prediction system 114.

Then, the facility value estimation device 10 executes the facility value estimation process, similar to the disaster recovery support process, and calculates the infrastructure impact degree (s1502).

The facility value estimation device 10 calculates an integrated risk assessment value by multiplying the damage probability obtained in s1501 by the infrastructure impact degree calculated in s1502 on a pipeline basis, and also creates integrated risk assessment information (s1503). The facility value estimation device 10 transmits the created integrated risk assessment information to the repair and restoration plan creation system 115, which creates maintenance plan information based on this information and displays it on the screen. This allows the user to create an annual maintenance plan for normal times, for example, so as to reduce the integrated risk of the pipeline.

(Integrated risk assessment information)
28 is a diagram showing an example of the integrated risk assessment information 1505. The integrated risk assessment information 1505 has data items of an ID 15051 of each pipeline, a failure probability 15052 of each pipeline, an infrastructure impact degree 15053 of each pipeline, and an integrated risk assessment value 15054 of each pipeline.

As described above, the facility value estimation device 10 of this embodiment calculates the facility value and its calculation accuracy for each facility, and if there is a low-accuracy facility with a low calculation accuracy of the facility value, it identifies a group (community) consisting of multiple facilities including the low-accuracy facility and other communities that are highly similar in specified elements, recalculates the facility value of the low-accuracy facility based on the facility value of the identified community, and displays facility value information for each of these facilities.

In other words, the facility value estimation device 10 of this embodiment calculates the facility value of each facility, and if the estimation accuracy of the facility value of a certain facility is low, it re-estimates the facility value based on the facility values of facilities in other communities.

In this way, the facility value estimation device 10 of this embodiment can stably calculate highly accurate facility values. In other words, since the facility value estimation device 10 quantitatively estimates the value of each facility, it can calculate facility values with higher accuracy than the conventional method in which facility values were quantified for each type of facility. This also greatly reduces the effort required of the user to modify the calculated facility value.

The present invention is not limited to the embodiments described above, and includes various modifications. The above-described embodiments have been described in detail to provide a better understanding of the present invention, and are not necessarily limited to those having all of the configurations described.

For example, some of the functions provided by each device in this embodiment may be provided in another device, or functions provided by another device may be provided in the same device.

In addition, in this embodiment, the facility value is the sum of the facility's human value and economic value, but this is just one example, and other factors (public value, cultural value, etc.) may also be taken into consideration.

In addition, in this embodiment, the edge weights 6024 in the network structure table 602 are based on the flow of people between facilities, but they may also be calculated based on other factors (for example, attributes of each facility as defined in the city data 901, or attributes of the area).

In addition, in this embodiment, the infrastructure is assumed to be a water pipeline, but the public facility management system 1 can be applied to other infrastructure. For example, by making the infrastructure data 111 information on the usage of the target facility and infrastructure and the network structure of the infrastructure, the public facility management system 1 can be applied to other infrastructure.

For example, when the infrastructure is a road, a traffic volume table and a road network structure table are provided as infrastructure data 111. The traffic volume table is information indicating the traffic volume for each road, and includes the road ID, road type, average traffic volume, etc. The road network structure table is information indicating the graph network structure of the road, and includes the road ID, start node ID, end node ID, latitude or longitude of the start and end points, etc. Here, a node is a connection between roads (a branch of a road).

Furthermore, when the infrastructure is a power grid, a power demand table and a power facility network structure table are provided as infrastructure data 111. The power demand table is information indicating the demand of the facility to which power is supplied, and includes the facility ID, average power usage, or average power usage fee. The power network structure table is information indicating the power network structure, and includes the edge ID, starting point power facility ID, ending point power facility ID, latitude or longitude of the starting point and ending point power facilities, etc. Here, an edge is a power transmission route such as an electric wire connecting power facilities (nodes).

The above description of this specification makes at least the following clear. That is, the re-estimation unit of the facility value estimation device 10 of this embodiment may accept input of corrections to the recalculated evaluation value of the low-accuracy facility from the user.

In other words, by accepting input from the user to correct the recalculated evaluation value of a low-accuracy facility, a more accurate facility value can be set. For example, if data necessary for calculating or recalculating the facility value is missing, particularly when a trained model is used to calculate or recalculate the facility value and there is variation (noise) in the teacher label, the calculation accuracy of the facility value tends to be low. In such a case, the user can ultimately correct the facility value. Note that since the facility value estimation device 10 calculates the estimation accuracy of the facility value (e.g., the probability distribution of variance), the calculation result is quantified, allowing the user to perform highly accurate verification.

The estimation unit of the facility value estimation device 10 of this embodiment may also calculate the evaluation value of each facility and the calculation accuracy of the evaluation value by inputting attribute information of each facility and the area to which the facility belongs into a first trained model in which attribute information of the facility and the area to which the facility belongs is used as an input value and the evaluation value of the facility is used as an output value.

In this way, by calculating the facility value using the first analysis model, which uses at least the city data 109 as an input value and the facility value as an output value, the facility value in a specified area, etc. can be calculated rationally.

In addition, when the re-estimation unit of the facility value estimation device 10 of this embodiment determines that there is a low-accuracy facility, it may recalculate the evaluation value of the low-accuracy facility based on the evaluation values of facilities in another group of multiple facilities that have human connections that are highly similar to the human connections between facilities in the group of multiple facilities that includes the low-accuracy facility.

In this way, for low-accuracy facilities, the evaluation value of the low-accuracy facility is recalculated based on the facility values of other communities (particularly communities in areas other than the designated area) that have similar patterns of human connections (e.g., people flow) to the community to which the low-accuracy facility belongs. In this way, if the accuracy of the previously calculated facility value was low, it is possible to recalculate the facility value with higher calculation accuracy.

In other words, in each community consisting of multiple facilities, even if the areas to which they belong are different, if the types of human connections between the facilities are similar, the facility value of each facility within that community is likely to be similar. For example, in any community, facilities with a high flow of people to surrounding facilities (e.g., hospitals) tend to be important facilities and have high facility values. If we focus on this property, there is a high possibility that we can recalculate the facility value with greater accuracy by performing the above-mentioned recalculation.

In addition, when the re-estimation unit of the facility value estimation device 10 of this embodiment determines that there is a low-accuracy facility, it may recalculate the evaluation value of the low-accuracy facility by inputting information on the human connections between facilities in an area that includes the low-accuracy facility into a second trained model that uses information on the human connections between facilities in a specified area as an input value and the evaluation value of the facility in the specified area as an output value.

In this way, when re-estimating facility value, the facility value is recalculated using a second analysis model that uses at least the human connections between facilities in a certain area (such as edge weights corresponding to people flow) as input values and the facility value as output values, making it possible to accurately calculate facility value in a specified area, etc.

Furthermore, the estimation unit of the facility value estimation device 10 of this embodiment may calculate an evaluation value of the facility based on a parameter indicating the human value of the facility, a parameter indicating the economic value of the facility, and a parameter indicating the relative importance of the human and economic aspects in the area to which the facility belongs.

In this way, by calculating the facility value using human value, economic value, and weighting coefficients (a, b), etc., the facility value of the facility can be calculated rationally taking into account social attributes, etc.

The estimation unit of the facility value estimation device 10 of this embodiment may also calculate an evaluation value of the facility based on at least one of the name, type, coordinates, area, capacity, number of employees, number of hospital beds, sales, or published price of at least one of the facilities, such as a hospital, school, childcare facility, nursing care facility, evacuation shelter, house, office building, factory, or industrial facility, and the demand or layout of public facilities in the area to which the facility belongs.

Calculating facility value based on this information makes it possible to calculate facility value that reflects the actual situation in society.

The display unit of the facility value estimation device 10 of this embodiment may display information on facilities for which the calculation accuracy of the evaluation value is less than the predetermined threshold in a first display format, and display information on the other facilities in a second display format.

In this way, by using different display formats depending on whether the facility is low-precision or not, users can easily recognize the need to revise or recalculate the facility value.

The public facility impact assessment system 1 of this embodiment may also include a public facility management unit that acquires information on abnormal parts of the public facility from a specified information processing system, and calculates an assessment value indicating the degree of need to repair the abnormal parts based on information on the location of the abnormal parts and the calculated infrastructure impact degree.

In this way, by creating recovery plan information based on information on abnormal parts of the infrastructure obtained from the sensor status detection system 113, etc., and the degree of impact on the infrastructure, it is possible to formulate an appropriate infrastructure recovery plan according to the current state of each facility in the region, etc.

The public facility impact assessment system 1 of this embodiment may also include a public facility management unit that acquires information on the risk of damage to the public facility from a specified information processing system, and calculates an assessment value indicating the degree of need to respond to the risk of damage based on the calculated infrastructure impact and the acquired information on the risk of damage.

In this way, by creating integrated risk assessment information 1505 based on information on infrastructure damage obtained from the damage risk prediction system 114, etc., and the impact on infrastructure, it is possible to formulate an appropriate infrastructure maintenance plan according to the current state of each facility in the region, etc.

1 Public facility impact assessment system, 10 Facility value estimation device, 20 Data management device, 30
Facility value estimation system, 113 Sensor condition detection system, 114 Damage risk prediction system, 115 Repair and restoration planning system

Claims (9)

A processor and a memory,
an estimation unit that reads attribute information of each facility belonging to the designated area and attribute information of the designated area from the memory that stores attribute information of the designated area, attribute information of each facility belonging to the designated area, and people flow data that indicates time-series movement of people in the designated area, and calculates an evaluation value of each facility belonging to the designated area and a calculation accuracy that is the reciprocal of the variance of the evaluation value by inputting the attribute information of each facility belonging to the designated area and the attribute information of the designated area into a first trained model that uses the attribute information of the facility and the attribute information of the area to which the facility belongs as an input value and an evaluation value of the facility as an output value ;
a re-estimation unit that determines whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, thereby determining whether the designated area includes a low-accuracy facility, which is a facility whose calculation accuracy of the evaluation value is less than a predetermined threshold, and when it is determined that the low-accuracy facility is included in the designated area, recalculates the evaluation value of the low-accuracy facility by inputting information on human connections between facilities in the designated area based on people flow data in the designated area read from the memory into a second trained model in which information on human connections between facilities in the designated area is used as an input value and the evaluation value of the facility in the designated area is used as an output value ;
a display unit that displays information on the calculated or recalculated evaluation value of the facility; and
Equipped with
The second trained model is created by a learning process in which information on human connections between facilities in areas other than the designated area is used as training data for input values, and evaluation values of facilities in the other areas is used as training data for output values.
Facility value estimator. The facility value estimation device according to claim 1, wherein the re-estimation unit receives input of corrections to the recalculated evaluation value of the low-accuracy facility from a user. The second trained model uses information on human connections obtained by reconstructing each facility belonging to the specified area into a plurality of groups by extracting a densely connected subset from the dependency relationships between facilities based on human connections in the specified area as an input value, and uses evaluation values of facilities in the specified area as an output value,
When it is determined that there is a low-accuracy facility, the re-estimation unit creates information on the human connections in which the facilities belonging to the designated area are reconstructed into a plurality of groups by extracting a densely connected subset from the dependency relationships between facilities based on human connections in the designated area, and recalculates the evaluation value of the low-accuracy facility by inputting the created information on the human connections into the second trained model.
The facility value estimation device according to claim 1 . The attribute information of the facility includes a parameter indicating the human value of the facility and a parameter indicating the economic value of the facility;
The attribute information of an area includes parameters indicating the relative importance of human and economic aspects in the area;
The estimation unit is
Calculating the human value of each facility belonging to the designated area by inputting attribute information of each facility belonging to the designated area into a first human trained model having facility attribute information as an input value and the human value of the facility as an output value;
Calculating the economic value of each facility belonging to the designated area by inputting attribute information of each facility belonging to the designated area into a first economic trained model having attribute information of the facility as an input value and an economic value of the facility as an output value;
calculating an evaluation value of each facility belonging to the designated area as a linear sum of the human value and the economic value of each facility, using a parameter indicating the relative importance of the human aspect and the economic aspect in the designated area;
The facility value estimation device according to claim 1 . the estimation unit inputs attribute information of each facility belonging to the designated area, including at least one of the name, type, coordinates, area, capacity, number of employees, number of hospital beds, sales, and posted price of at least one of a hospital, a school, a childcare facility, a nursing home, an evacuation shelter, a house, an office building, a factory, and an industrial facility, and attribute information of the designated area into the first trained model, thereby calculating an evaluation value of each facility belonging to the designated area and a calculation accuracy which is the inverse of the variance of the evaluation value;
The facility value estimation device according to claim 1 . The facility value estimation device according to claim 1, wherein the display unit displays information on facilities among the facilities for which the calculation accuracy of the evaluation value is less than the predetermined threshold in a first display format, and displays information on the other facilities in a second display format. A processor and a memory,
an estimation unit that reads attribute information of each facility belonging to the designated area and attribute information of the designated area from the memory that stores attribute information of the designated area, attribute information of each facility belonging to the designated area, and people flow data that indicates time-series movement of people in the designated area, and calculates an evaluation value of each facility belonging to the designated area and a calculation accuracy that is the reciprocal of the variance of the evaluation value by inputting the attribute information of each facility belonging to the designated area and the attribute information of the designated area into a first trained model that uses the attribute information of the facility and the attribute information of the area to which the facility belongs as an input value and an evaluation value of the facility as an output value ;
a re-estimation unit that determines whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, thereby determining whether the designated area includes a low-accuracy facility, which is a facility whose calculation accuracy of the evaluation value is less than a predetermined threshold, and when it is determined that the low-accuracy facility is included in the designated area, recalculates the evaluation value of the low-accuracy facility by inputting information on human connections between facilities in the designated area based on people flow data in the designated area read from the memory into a second trained model in which information on human connections between facilities in the designated area is used as an input value and the evaluation value of the facility in the designated area is used as an output value ;
an evaluation unit that calculates an infrastructure impact degree that indicates the magnitude of the impact of the public facility on each facility based on information on the location of the public facility, the calculated or recalculated facility value of each facility, and information on the location of each facility;
a display unit that displays information on the calculated or recalculated evaluation value of the facility and the calculated impact degree on infrastructure;
Equipped with
The second trained model is created by a learning process in which information on human connections between facilities in areas other than the designated area is used as training data for input values, and evaluation values of facilities in the other areas is used as training data for output values.
Public Facilities Impact Assessment System. The public facility impact assessment system of claim 7, further comprising a public facility management unit that acquires information on the risk of damage to the public facility from a specified information processing system, and calculates an evaluation value indicating the degree of need to respond to the risk of damage based on the calculated infrastructure impact degree and the acquired information on the risk of damage. An information processing device,
an estimation process of reading attribute information of each facility belonging to the designated area and attribute information of the designated area from a memory that stores attribute information of the designated area, attribute information of each facility belonging to the designated area, and people flow data indicating time-series movement of people in the designated area, and inputting the attribute information of each facility belonging to the designated area and the attribute information of the designated area into a first trained model that uses the attribute information of the facility and the attribute information of the area to which the facility belongs as input values and the evaluation value of the facility as an output value, thereby calculating an evaluation value of each facility belonging to the designated area and a calculation accuracy that is the inverse of the variance of the evaluation value ;
a re-estimation process for determining whether the calculation accuracy of the evaluation value of each facility belonging to the designated area is equal to or greater than a predetermined threshold, thereby determining whether the designated area includes a low-accuracy facility, which is a facility whose calculation accuracy of the evaluation value is less than a predetermined threshold, and, if it is determined that the low-accuracy facility is included in the designated area, recalculating the evaluation value of the low-accuracy facility by inputting information on human connections between facilities in the designated area based on people flow data in the designated area read from the memory into a second trained model in which information on human connections between facilities in the designated area is used as an input value and the evaluation value of the facility in the designated area is used as an output value ;
An evaluation process for calculating an infrastructure impact degree that indicates the magnitude of the impact of the public facility on each facility based on information on the location of the public facility, the calculated or recalculated facility value of each facility, and information on the location of each facility;
a display process for displaying information on the calculated or recalculated evaluation value of the facility and the calculated impact degree on infrastructure;
Run
The second trained model is created by a learning process in which information on human connections between facilities in areas other than the designated area is used as training data for input values, and evaluation values of facilities in the other areas is used as training data for output values.
Public facility impact assessment methods.

Images