JP7726019B2 - Salt damage occurrence determination system, worker terminal, and salt damage occurrence determination method - Google Patents

Description

The present invention relates to a salt damage occurrence determination system, a worker terminal, a model learning device, and a salt damage occurrence determination method.

Patent Document 1 describes an insulator contamination detection device designed to automatically detect insulator contamination and determine the optimal time to clean the insulators. The insulator contamination detection device includes a waveform recorder attached to a transmission line tower, a sensor that measures the waveform of insulator noise, a waveform comparison circuit that compares the waveform with that of uncontaminated insulators, a CPU that outputs a transmission signal when the waveform is abnormal, and a transmitter that transmits the waveform; and an analyzer installed in a monitoring station, which includes a receiver that receives the waveform, a noise comparison circuit that compares noise at a specific frequency within the waveform with that of uncontaminated insulators, a CPU that outputs a display signal when the comparison difference is equal to or greater than a predetermined value, and a display that displays the comparison results. The insulator contamination detection device measures the waveform of insulator noise, compares the noise at a specific frequency with that of uncontaminated insulators, and automatically reports insulator contamination when the comparison difference is equal to or greater than a predetermined value.

Japanese Patent Application Publication No. 10-197326

Power transmission and distribution equipment such as insulators installed on utility poles and transmission line towers can experience partial discharges due to salt damage (deterioration of insulation), which can cause noise (corona noise) and radio interference in the neighborhood. Furthermore, as salt damage progresses, it can cause widespread power outages. For this reason, those responsible for managing power equipment, such as power transmission companies, regularly patrol areas where salt damage is likely to occur and perform inspections and maintenance (cleaning, etc.) of power transmission and distribution equipment.

However, in reality, identifying areas and facilities affected by salt damage relies on the intuition and experience of workers, and if there is even the slightest bit of uncertainty, workers go to the site to carry out inspections, maintenance, and other work. In particular, when major natural disasters such as typhoons, strong winds, or heavy snowfall occur, the scope of inspections and maintenance must be expanded, and workers spend a great deal of time and effort taking measures and responding to salt damage.

The present invention has been made in consideration of the above background, and aims to provide a salt damage occurrence determination system, an operator terminal , and a salt damage occurrence determination method that can easily and accurately determine the presence or absence of salt damage in power transmission and distribution equipment.

One of the present inventions for achieving the above-mentioned objectives is an information processing system (salt damage occurrence determination system) that determines whether salt damage has occurred in power transmission and distribution equipment. It stores an analytical model, which is a machine learning model trained using training data configured with information based on sound data acquired at the site where the power transmission and distribution equipment is installed as an explanatory variable and information regarding leak sounds emitted from the power transmission and distribution equipment due to salt damage as a target variable, and outputs information based on the target variable obtained by inputting information based on the sound data acquired at the site as the explanatory variable into the analytical model.

The salt damage determination system of the present invention automatically outputs information about leak sounds emitted from power transmission and distribution equipment located near the site based on sound data acquired on-site. Therefore, workers performing on-site inspection and maintenance of power transmission and distribution equipment can easily determine whether salt damage has occurred in the power transmission and distribution equipment located near the site, allowing for efficient inspection and maintenance work. Furthermore, by using an analytical model trained using sound data acquired in a variety of environments, workers can accurately determine whether salt damage has occurred.

Another aspect of the present invention for achieving the above-mentioned objectives is an information processing system (salt damage occurrence determination system) that determines whether salt damage has occurred in power transmission and distribution equipment. It stores an analytical model, which is a machine learning model trained using training data configured with information based on sound data and environmental measurement information acquired at the site where the power transmission and distribution equipment is installed as explanatory variables, and information on leak noise emitted from the power transmission and distribution equipment due to salt damage as a target variable. The analytical model inputs information based on sound data and environmental measurement information acquired at the site as the explanatory variables, and outputs information based on the target variable.

The salt damage determination system of the present invention automatically outputs information about leak sounds emitted from power transmission and distribution equipment located near the site based on sound data and environmental measurement information (temperature, humidity, air pressure, etc.) acquired on-site. This allows workers inspecting and maintaining power transmission and distribution equipment at the site to easily determine whether salt damage has occurred in the power transmission and distribution equipment located near the site, allowing for efficient inspection and maintenance work. Furthermore, by using an analytical model trained using sound data and environmental measurement information as explanatory variables, information about leak sounds is generated taking into account the impact of differences in the on-site environment (temperature, humidity, air pressure, etc.) on the sound data, allowing workers to accurately determine whether salt damage has occurred.

Another aspect of the present invention for achieving the above objective is the salt damage occurrence determination system, which stores equipment information relating to the power transmission and distribution equipment being managed, and reflects the contents of the objective variables obtained by inputting information based on the sound data acquired on-site as the subject of analysis into the analysis model.

The salt damage occurrence determination system of the present invention automatically reflects the contents of the objective variable obtained by inputting explanatory variables based on information on the analysis target obtained on-site into an analytical model in the facility information managed for power transmission and distribution facilities. This allows those responsible for managing power transmission and distribution facilities at organizations such as electric utilities to manage their facilities more efficiently.

Another aspect of the present invention for achieving the above objective is an information processing device (worker terminal) that constitutes the above-mentioned salt damage occurrence determination system. It is used by workers working at sites where power transmission and distribution facilities are located, stores the analytical model, and outputs information based on the objective variables obtained by inputting information based on sound data acquired at the site as the analysis target into the analytical model as explanatory variables.

In this way, the worker terminal of the present invention outputs information based on the objective variables obtained by inputting sound data acquired on-site into an analysis model, allowing workers to check the contents of the objective variables on-site and proceed with work more efficiently.

Another aspect of the present invention for achieving the above objective is an information processing device (model learning device) that constitutes the above-mentioned salt damage occurrence determination system, which is communicatively connected to a worker terminal used by a worker performing work near a site where power transmission and distribution equipment is located, generates the analytical model as needed, and transmits the generated analytical model to the worker terminal.

In this way, the model learning device of the present invention generates trained analytical models as needed and transmits them to worker terminals, allowing for efficient management of the analytical models on the worker terminals used by each worker, keeping them up to date.

Furthermore, the problems disclosed in this application and the solutions thereto will be made clear in the detailed description and drawings.

The present invention makes it possible to easily and accurately determine whether or not salt damage has occurred in power transmission and distribution equipment. This can, for example, assist workers in the work of inspections and maintenance.

1 is a diagram showing a schematic configuration of a salt damage occurrence determination system. FIG. 10 is a diagram showing a situation in which a worker at a site acquires information (sound data (environmental sound), measurement information) used to generate explanatory variables. FIG. 2 is a diagram illustrating main functions of the model learning device. FIG. 2 is a diagram illustrating a hardware configuration of a model learning device. FIG. 2 is a diagram illustrating an example of sound data. FIG. 2 is a diagram illustrating main functions of a worker terminal. FIG. 2 is a diagram illustrating a hardware configuration of a worker terminal. 1 is an example of a data structure of learning data. 10 is an example of a data structure of facility information. 10 is an example of a data structure of worker terminal information. 10 is an example of a data structure of an analysis result. 10 is a flowchart illustrating a learning process. 10 is a flowchart illustrating an analysis process. 10 is a flowchart illustrating an analysis model receiving process. 10 is a flowchart illustrating a facility information update process.

Embodiments of the present invention will be described below, with appropriate reference to the drawings. Note that the following description and drawings are examples for explaining the present invention, and have been omitted or simplified as appropriate for clarity of explanation. The present invention can also be implemented in various other forms. Unless otherwise specified, each component may be singular or plural.

In the following explanation, identical or similar components may be assigned the same reference numerals, and duplicate explanations may be omitted. Also, in the following explanation, the letter "S" placed before a reference numeral indicates a processing step.

Figure 1 shows the general configuration of an information processing system (hereinafter referred to as "salt damage occurrence determination system 1") described as one embodiment of the present invention. As shown in the figure, salt damage occurrence determination system 1 includes a model learning device 100, which is an information processing device (computer) installed in a facility management department or data center operated by an electric utility, and one or more worker terminals 200, which are information processing devices used by workers 2 who perform work such as inspection and maintenance of power transmission and distribution equipment on-site. The worker terminals 200 are, for example, smartphones, tablets, notebook computers, dedicated terminals, etc.

The model learning device 100 and the worker terminal 200 are communicatively connected via a communication network 5. The communication network 5 is, for example, a LAN (Local Area Network), a WAN (Wide Area Network), the Internet, various wireless communication networks (3G communication networks, 4G communication networks, 5G communication networks, etc.), and various dedicated lines.

The salt damage occurrence determination system 1 uses a machine learning model (hereinafter referred to as the "analysis model") to analyze sound data (environmental sounds) acquired around sites where various types of power transmission and distribution equipment (insulators, power transmission and distribution lines, various substation equipment, various power distribution equipment, etc.) are installed on utility poles, power transmission line towers, substations, etc., to determine the presence or absence of sounds (hereinafter referred to as "leak sounds") emitted from the power transmission and distribution equipment due to salt damage, and provides information based on the determination results as analysis results to users such as workers 2 working on site and managers of the power transmission and distribution equipment.

During normal inspection/maintenance, or when there is a high possibility that salt damage has occurred to the power transmission/distribution equipment due to a natural disaster (typhoon, strong wind, rainfall, snowfall, fog, etc.), worker 2 goes to the site where the power transmission/distribution equipment is located, refers to the analysis results output by worker terminal 200, determines whether salt damage has occurred in the power transmission/distribution equipment around the site, and takes measures such as cleaning if necessary.

In this way, the salt damage occurrence determination system 1 automatically generates analysis results based on information acquired on-site and provides them to the user, allowing the user to easily and accurately determine whether salt damage has occurred based on the analysis results provided, without relying on intuition or experience. Furthermore, it is now possible to accurately detect subtle leak sounds that would be undetectable to the human ear or leak sounds buried in noise in power transmission and distribution equipment, allowing users to accurately determine whether salt damage has occurred and efficiently perform inspection and maintenance work on power transmission and distribution equipment.

The model learning device 100 learns an analytical model using learning data (teacher data, training data) that associates explanatory variables (features) based on information acquired at the site (sound data (environmental sounds), measurement information) with objective variables (labels) received from the user (presence or absence of leak sounds, type of leak sound, etc.). The information used to generate the explanatory variables is acquired under various on-site environments. The objective variables (labels) may be, for example, the results of determining whether the sound data contains components derived from salt damage using frequency analysis based on Fourier transform or environmental sound identification technology (acoustic event analysis (classification) technology, acoustic scene classification (identification) technology, abnormal sound detection technology, etc.). Note that a machine learning model used to identify environmental sounds may be used as the analytical model. Furthermore, for example, an existing environmental sound corpus may be used as the sound data (explanatory variables) for normal cases (not including leak sounds).

Figure 2 shows a worker 2 acquiring information (sound data (environmental sounds), measurement information) used to generate explanatory variables at a site where power transmission and distribution equipment 3 is installed. Worker 2 sets up a tripod at the site to which a microphone 27 and measurement device 28 (environmental sensors (temperature sensor, humidity sensor, barometric pressure sensor, etc.)) are attached to acquire (collect) sound data and measurement information. In this example, worker 2 points microphone 27 toward an insulator installed at the top of a utility pole. The information acquired by microphone 27 and measurement device 28 is input, for example, into a worker terminal 200 (e.g., a notebook personal computer, smartphone, tablet, etc.) that worker 2 has brought to the site.

Furthermore, by using a highly sensitive microphone 27, sound data can be acquired from a wider range of power transmission and distribution equipment 3. In this example, the microphone 27 and measuring device 28 are connected to a tripod installed on the ground to acquire information (sound data, measurement information), but these may also be attached to a vehicle used by worker 2 for inspection and maintenance. This saves worker 2 the trouble of setting up and removing the device, allowing for more efficient work.

Figure 3A shows the main functions of the model learning device 100. As shown in the figure, the model learning device 100 has the following functions: a memory unit 110, an information acquisition management unit 120, a learning data generation unit 130, an analysis model learning unit 135, an analysis model distribution unit 140, an analysis result receiving unit 145, and an equipment information update unit 150.

Of the above functions, the memory unit 110 stores sound data 111, measurement information 112, learning data 113, analysis model 114, equipment information 115, and worker terminal information 116.

Of these, sound data 111 includes a large amount of sound data (sound waveform data) acquired at various sites. Note that, because the leak sound emitted from power transmission and distribution equipment in the event of salt damage is typically continuous, a single piece of sound data is acquired as, for example, data of the length of time required to accurately determine whether or not a leak sound is present (hereinafter referred to as a "sound clip/audio clip").

Figure 4 shows an example of sound data. The sound data is, for example, waveform data for a predetermined period (such as tens of milliseconds to tens of seconds) that serves as the unit of analysis.

Returning to Figure 3A, the measurement information 112 is data measured using various sensors (environmental sensors (temperature, humidity, air pressure, etc.), position sensors (GPS sensors (GPS: Global Positioning System), etc.)) at the site where the sound data 111 is acquired using measuring equipment such as the measuring device 28 described below. The measurement information 112 is used, for example, to enable the analysis model 114 to learn the effect on the waveform of the sound data of differences in the environment at the site where the sound data was acquired (differences in temperature, humidity, air pressure, etc.).

The training data 113 is training data used to train the analysis model 114. The training data is data in which explanatory variables generated based on information (sound data, measurement information) acquired on-site are associated with objective variables (labels) (presence or absence of leak sound, type of leak sound, etc.). The explanatory variables are, for example, features extracted from sound clips. Note that the training data 113 only needs to include at least sound data as explanatory variables, and whether or not measurement information is used as an explanatory variable, and if so, what type of measurement information to use, can be determined according to need and the situation.

The analytical model 114 is a machine learning model trained using the training data 113. Features of the sound data 111 used in the analytical model 114 include, for example, Mel frequency cepstrum coefficients (MFCCs), cepstrums, MPEG-7 acoustic features, and other acoustic features. The analytical model 114 is, for example, a polynomial, a determinant, a mathematical expression, a vector, or the like, including adjustable parameters. The type of analytical model 114 is not necessarily limited, but examples include various deep learning models (CNNs: convolutional neural networks), RNNs (recurrent neural networks), Gaussian mixture models (GMMs), hidden Markov models (HMMs), and support vector machines (SVMs). The analytical model 114 may also be a so-called anomaly detection model that learns abnormal states by learning steady states (e.g., a machine learning model trained using an autoencoder (VAE (variational autoencoder)), etc.).

Facility information 115 is information used by electric utilities and other entities to manage power transmission and distribution facilities, and includes information about the power transmission and distribution facilities installed in the area (such as the type and model of the facility, the installation location of the facility, the date of installation, the dates and details of past inspections/maintenance work, and whether or not there is current salt damage).

The worker terminal information 116 includes information about the worker terminal 200 to which the analytical model 114 is delivered from the model learning device 100.

Among the functions shown in the same figure, the information acquisition management unit 120 manages sound data 111 and measurement information 112 acquired at the site, in a test environment, etc.

The training data generation unit 130 generates the training data 113. The training data generation unit 130 generates the training data 113, for example, by presenting explanatory variables and accepting from the user, via a user interface, the objective variables to be set for the explanatory variables.

The analytical model learning unit 135 uses the learning data 113 to learn the analytical model 114.

The analytical model distribution unit 140 distributes (transmits) the latest analytical model 114 learned by the analytical model learning unit 135 to the worker terminal 200 via the communication network 5.

The analysis result receiving unit 145 receives the analysis results sent from the worker terminal 200 via the communication network 5.

The equipment information update unit 150 receives the analysis results sent from the worker terminal 200 and reflects the contents of the received analysis results in the equipment information 115.

Figure 3B is a diagram showing the hardware configuration of the model learning device 100. As shown in the figure, the model learning device 100 includes a processor 11, a main memory device 12, an auxiliary memory device 13, an input device 14, an output device 15, and a communication device 16. The model learning device 100 may be configured from multiple information processing devices connected to each other so that they can communicate with each other.

All or part of the configuration shown in the figure may be realized using virtual information processing resources provided using virtualization technology, process space separation technology, or the like, such as virtual servers provided by a cloud system. All or part of the functions of model learning device 100 may be realized, for example, by services provided by a cloud system via an API (Application Programming Interface), etc. All or part of the functions of model learning device 100 may be realized using, for example, SaaS (Software as a Service), PaaS (Platform as a Service), IaaS (Infrastructure as a Service), etc.

In the same diagram, the processor 11 is configured using, for example, a CPU (Central Processing Unit), MPU (Micro Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), AI (Artificial Intelligence) chip, etc.

The main memory device 12 is a device used by the processor 11 when executing programs, and may be, for example, a read-only memory (ROM), a random access memory (RAM), or a non-volatile memory (NVRAM (Non-Volatile RAM)). The various functions of the model learning device 100 are realized when the processor 11 reads programs and data stored in the auxiliary memory device 13 into the main memory device 12 and executes them.

The auxiliary storage device 13 may be, for example, an SSD (Solid State Drive), a hard disk drive, an optical storage device (CD (Compact Disc), DVD (Digital Versatile Disc), etc.), a storage system, an IC card, a read/write device for recording media such as an SD card or optical recording media, or a storage area of a cloud server. Programs and data can be read into the auxiliary storage device 13 via a recording media reader or communication device 16. Programs and data stored in the auxiliary storage device 13 are read into the main storage device 12 as needed.

The input device 14 is an interface that accepts input from outside, and may be, for example, a touch panel, keyboard, mouse, card reader, pen-input tablet, or voice input device.

The output device 15 is an interface that outputs various information such as processing progress and results. The output device 15 is, for example, a display device (liquid crystal monitor, LCD (Liquid Crystal Display), graphics card, etc.) that visualizes the various information described above, a device that converts the various information described above into audio (audio output device (speaker, etc.)), or a device that converts the various information described above into text (printer, etc.). Note that, for example, the model learning device 100 may be configured to input and output information to and from other devices via the communication device 16.

The input device 14 and output device 15 form a user interface that accepts and presents information to the user.

The communication device 16 is a device that enables communication with other devices. The communication device 16 is a wired or wireless communication interface that enables communication with other devices via a communication network 5 such as the Internet, and is, for example, a NIC (Network Interface Card), a wireless communication module, a USB module, etc.

The model learning device 100 may be equipped with, for example, an operating system, a file system, a DBMS (Database Management System) (relational database, NoSQL, etc.), a KVS (Key-Value Store), etc.

The above-mentioned functions of the model learning device 100 are realized by the processor 11 reading and executing programs stored in the main memory device 12, or by the hardware (FPGA, ASIC, AI chip, etc.) that constitutes the model learning device 100. The model learning device 100 stores the above-mentioned various information (data), for example, as database tables or files managed by a file system.

Figure 5A shows the main functions of the worker terminal 200. As shown in the figure, the worker terminal 200 has the following functions: a memory unit 210, an analysis target information acquisition unit 220, an explanatory variable generation unit 225, an analysis processing unit 230, an analysis result presentation unit 235, an analysis result transmission unit 240, and an analysis model update unit 245.

Of the above functions, the memory unit 210 stores the following information: analysis target information 211, explanatory variables 212, analysis model 114, and analysis results 213.

Of the above information, analysis target information 211 is information (sound data, measurement information) input at the site as the target for analysis.

Explanatory variables 212 are explanatory variables (features) generated based on the information input as the subject of analysis (sound data input from microphone 27, measurement information input from measurement device 28).

The analytical model 114 is distributed from the model learning device 100. The analytical model 114 is the same as described above, so further explanation will be omitted.

The analysis results 213 include information about the analysis results performed by the analysis processing unit 230 using the analysis model 114 (presence or absence of leak noise, type of leak noise, presence or absence of salt damage, etc.).

Of the above functions, the explanatory variable generation unit 225 extracts features from the analysis target information 211 and generates explanatory variables 212.

The analysis processing unit 230 inputs the explanatory variables 212 into the analysis model 114, thereby obtaining the objective variables output by the analysis model 114.

The analysis result presentation unit 235 generates the analysis result 213 based on the objective variable acquired by the analysis processing unit 230 and presents the generated analysis result 213 to the user.

The analysis result transmission unit 240 transmits the analysis result 213 to the model learning device 100 via the communication network 5.

The analytical model update unit 245 receives the analytical model 114 sent from the model learning device 100 via the communication network 5, and updates the analytical model 114 stored in the memory unit 210 to the received analytical model 114. The analytical model update unit 245 may also receive only the update difference of the analytical model 114 from the model learning device 100, and update the analytical model 114 using the received update difference.

Figure 5B shows an example of the hardware configuration of the worker terminal 200. As shown in the figure, the worker terminal 200 includes a processor 21, a main memory device 22, an auxiliary memory device 23, an input device 24, an output device 25, a communication device 26, a microphone 27, and a measurement device 28. Of these, the processor 21, main memory device 22, auxiliary memory device 23, input device 24, output device 25, and communication device 26 are similar to the processor 11, main memory device 12, auxiliary memory device 13, input device 14, output device 15, and communication device 16 of the model learning device 100, respectively, and therefore will not be described here. The microphone 27 and measurement device 28 may be integrated into the worker terminal 200 itself or may be external.

The microphone 27 is a device that acquires sound data (sound waveform data) and may be, for example, a dynamic microphone or a condenser microphone. The microphone 27 is equipped with an amplifier circuit, which amplifies the acquired sound data to a specified required power level.

The measuring device 28 is composed of various sensors (temperature sensor, humidity sensor, air pressure sensor, etc.) and amplifier circuits for the sensor output values, and outputs measured values related to the on-site environment (temperature, humidity, air pressure, etc.).

Figure 6 shows an example of the data structure of the learning data 113. As shown in the figure, the example learning data 113 is composed of one or more records each having items such as a data ID 611, sound data 612, measurement information 613, presence or absence of leak sound 614, and type of leak sound 615. One record of the learning data 113 corresponds to one piece of learning data.

Of the above items, data ID 611 stores the data ID, which is the identifier of the learning data. Sound data 612 and measurement information 613 correspond to explanatory variables. Furthermore, presence/absence of leak sound 614 and type of leak sound 615 correspond to objective variables. For example, the user checks the contents of the explanatory variables presented on the screen via a user interface provided by the model learning device 100, and sets the contents of the objective variables corresponding to those explanatory variables.

Figure 7 shows an example of the data structure of equipment information 115. As shown in the figure, the illustrated equipment information 115 is composed of one or more records each having fields such as equipment ID 711, type 712, model 713, installation location 714, installation date 715, previous work date 716, work content 717, and salt damage 718. One record in the equipment information 115 corresponds to one power transmission and distribution facility.

Of the above items, equipment ID 711 stores an identifier for the power transmission and distribution equipment (hereinafter referred to as "equipment ID"). Type 712 stores information indicating the type of power transmission and distribution equipment 3. Model 713 stores information indicating the model (model number) of the power transmission and distribution equipment 3. Installation location 714 stores information indicating the location (position) where the power transmission and distribution equipment is installed. Installation date 715 stores the date on which the power transmission and distribution equipment was installed. Last work date 716 stores the date on which worker 2 last (most recently) performed work (inspection, maintenance, etc.) on the power transmission and distribution equipment. Work content 717 stores information indicating the content of the work last performed by worker 2 on the power transmission and distribution equipment. Salt damage presence/absence 718 stores information indicating whether salt damage has occurred on the power transmission and distribution equipment ("yes" if salt damage has occurred, and "no" if no salt damage has occurred).

Figure 8 shows an example of the data structure of worker terminal information 116. As shown in the figure, the illustrated worker terminal information 116 is composed of one or more records each having items such as a terminal ID 811, a network address 812, a current version 813, and a last update date and time 814. One record in the worker terminal information 116 corresponds to one worker terminal 200.

The terminal ID 811 stores the identifier of the worker terminal 200 (hereinafter referred to as the "terminal ID"). The NW address 812 stores the network address (such as an IP address) assigned to the worker terminal 200. The current version 813 stores information indicating the current version of the analytical model 114 stored in the worker terminal 200. The last update date and time 814 stores the date and time of the most recent update of the analytical model 114 of the worker terminal 200.

Figure 9 shows an example of the data structure of the analysis results 213. As shown in the figure, the illustrated analysis results 213 are made up of one or more records each having items such as sound data acquisition date 911, sound data acquisition location 912, presence or absence of leak sound 913, leak sound type 914, and presence or absence of salt damage 915. One record in the analysis results 213 corresponds to the analysis results performed on one piece of sound data acquired at a certain site.

Sound data acquisition date 911 stores the date on which the sound data (sound data to be analyzed) was acquired. Sound data acquisition location 912 stores information indicating the location at which the sound data was acquired. Leak sound presence/absence 913 stores information indicating the presence or absence of leak sound among the objective variables output by analysis model 114. Leak sound type 914 stores information indicating the type of leak sound among the objective variables output by analysis model 114. Salt damage occurrence/absence 915 stores information indicating whether salt damage is currently occurring in the power transmission and distribution equipment at the site ("yes" if it has occurred, "no" if it has not occurred).

The analysis result presentation unit 235 of the worker terminal 200 determines that salt damage has occurred if, for example, a leak sound is occurring at the site or if the intensity of the leak sound is equal to or greater than a preset value, and sets the salt damage occurrence status 915 in the analysis result 213 to "Yes."

Next, we will explain the main processes performed by the salt damage occurrence determination system 1.

Figure 10 is a flowchart illustrating the processing performed by the analytical model learning unit 135 of the model learning device 100 when learning the analytical model 114 (hereinafter referred to as "learning processing S1000"). Below, learning processing S900 will be explained using this figure.

First, the analytical model learning unit 135 reads the learning data 113 from the auxiliary storage device 13 to the main storage device 12 (S1011).

Next, the analytical model learning unit 135 uses the learning data 113 to learn the analytical model 114 (S1012).

Next, the analytical model distribution unit 140 distributes (transmits) the latest trained analytical model 114 to each worker terminal 200 via the communication network 5. (S1013)

Note that since training the analytical model 114 requires a large amount of information processing resources, the training may be performed using, for example, an API (Application Programming Interface) for machine learning provided by the cloud system.

The analytical model learning unit 135 may also verify the prediction accuracy of the trained analytical model 114, for example. In this case, for example, the learning data is pre-classified into training data and verification data, and the training data is used for learning, and the verification data is used for verification.

Figure 11 is a flowchart illustrating the process (hereinafter referred to as "analysis process S1100") performed by the worker terminal 200 when analyzing information to be analyzed (sound data, measurement information) obtained on-site. The analysis process S1100 may be performed by the worker terminal 200 alone, or may be performed by the worker terminal 200 in cooperation with resources such as cloud services via the communication network 5. The analysis process S1100 is started, for example, when worker 2 performs a predetermined start operation on the worker terminal 200. The analysis process S1100 will be described below with reference to this figure.

First, the analysis target information acquisition unit 220 of the worker terminal 200 acquires the analysis target information 211 from the microphone 27 and the measuring device 28 (S1111-S1112). At this time, the analysis target information acquisition unit 220 may present (display, etc.) the acquired analysis target information 211 to the worker 2.

Next, the explanatory variable generation unit 225 of the worker terminal 200 extracts features from the acquired analysis target information 211 and generates explanatory variables 212 (S1113).

Next, the analysis processing unit 230 of the worker terminal 200 inputs the explanatory variables 212 into the analysis model 114 and obtains the objective variables output by the analysis model 114 (S1114).

Next, the analysis result presentation unit 235 of the worker terminal 200 presents information based on the acquired objective variables as analysis results (presence or absence of leak noise, type of leak noise, presence or absence of salt damage, etc.) (S1115).

Next, the analysis result transmission unit 240 of the worker terminal 200 transmits the analysis results to the model learning device 100 via the communication network 5 (S1116).

In this way, Worker 2 can use the worker terminal 200 to accurately determine whether salt damage has occurred in the power transmission and distribution equipment at the site, even if he or she does not have the necessary knowledge or experience. This allows for efficient and reliable inspection of the power transmission and distribution equipment, significantly reducing the workload of Worker 2.

Figure 12 is a flowchart explaining the processing performed by the worker terminal 200 when updating the analytical model 114 (hereinafter referred to as the "analysis model reception processing S1200"). Below, the analysis model reception processing S1200 will be explained using this figure.

The analytical model update unit 245 of the worker terminal 200 constantly monitors whether a new analytical model 114 (the latest analytical model 114 trained using the latest training data 113) or an update difference has been received from the model learning device 100 (S1211). The analytical model distribution unit 140 of the model learning device 100 distributes the analytical model to the worker terminal 200 at an appropriate time, such as when the analytical model 114 has been updated with new training data 113.

When the analytical model update unit 245 receives a new analytical model 114 or an update difference from the model learning device 100, it updates the analytical model 114 to the contents of the received analytical model 114 (or reflects the update difference in the analytical model 114) (S1212).

In this way, the analytical model 114 of the worker terminal 200 is always kept up to date by learning from the latest learning data 113, allowing worker 2 to determine whether salt damage is occurring continuously and stably at the site.

Figure 13 is a flowchart explaining the processing performed by the model learning device 100 when updating the facility information 115 (hereinafter referred to as the "facility information update processing S1300"). Below, the facility information update processing S1300 will be explained using this figure.

The equipment information update unit 150 of the model learning device 100 constantly monitors whether analysis results have been received from the worker terminal 200 (S1311).

When the equipment information update unit 150 receives new analysis results from the operator terminal 200, it reflects the contents of the received analysis results in the contents of the corresponding power transmission and distribution equipment in the equipment information 115 (such as salt damage presence/absence 718) (S1312).

In this way, the analysis results of the power transmission and distribution equipment installed in various locations are managed centrally as equipment information 115, allowing, for example, those who manage power transmission and distribution equipment at organizations such as electric power utilities to manage the equipment efficiently. Furthermore, the analysis results received from the operator terminal 200 may be used, for example, as a target variable (label) when generating training data 113 for the analysis model 114.

The above describes in detail an embodiment of the present invention, but the above description is provided to facilitate understanding of the present invention and does not limit the present invention. The present invention may be modified or improved without departing from its spirit, and the present invention naturally includes equivalents. For example, the above embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to those that include all of the described configurations. Furthermore, it is possible to add, delete, or replace part of the configuration of the above embodiment with other configurations.

1 Salt damage occurrence determination system, 2 Worker, 3 Power transmission and distribution equipment, 27 Microphone, 28 Measuring device, 5 Communication network, 100 Model learning device, 110 Memory unit, 111 Sound data, 112 Measurement information, 113 Learning data, 114 Analysis model, 115 Equipment information, 116 Worker terminal information, 120 Information acquisition and management unit, 130 Learning data generation unit, 135 Analysis model learning unit, 140 Analysis model distribution unit, 145 Analysis result receiving unit, 150 Equipment information update unit, 200 Worker terminal, 210 Memory unit, 211 Analysis target information, 212 Explanatory variable, 213 Analysis result, 220 Analysis target information acquisition unit, 225 Explanatory variable generation unit, 230 Analysis processing unit, 235 Analysis result presentation unit, 240 Analysis result transmission unit, 245 Analysis model update unit, S1000 Learning process, S1100 analysis process, S1200 analysis model reception process, S1300 equipment information update process

Claims (6)

An information processing system for determining whether salt damage has occurred in power transmission and distribution equipment,
An analytical model is stored, which is a machine learning model trained using learning data configured using information based on sound data, which is environmental sound acquired at the site where the power transmission and distribution equipment is installed, as explanatory variables, and information indicating the presence or absence of leak sounds emitted from the power transmission and distribution equipment due to salt damage and information indicating the type of leak sounds if the leak sounds are present,
and outputting information based on the objective variables obtained by inputting information based on sound data acquired at the site as an analysis target into the analysis model as the explanatory variables.
A system for determining whether salt damage has occurred. An information processing system for determining whether salt damage has occurred in power transmission and distribution equipment,
an analytical model that is a machine learning model trained using learning data configured using information based on sound data, which is environmental sound acquired at a site where the power transmission and distribution equipment is installed, and measurement information, which is information acquired by an environmental sensor at the site, as explanatory variables, and information indicating the presence or absence of leak sounds emitted from the power transmission and distribution equipment due to salt damage and information indicating the type of leak sounds if the leak sounds are present, as objective variables;
outputting information based on the objective variables obtained by inputting information based on sound data and environmental measurement information acquired at the site as the analysis target into the analysis model as the explanatory variables;
A system for determining whether salt damage has occurred. The salt damage occurrence determination system according to claim 1,
storing facility information that is information about the power transmission and distribution facility to be managed;
reflecting the content of the objective variable obtained by inputting information based on the sound data acquired at the site as an analysis target into the analysis model in the facility information;
A system for determining whether salt damage has occurred. An information processing device constituting the salt damage occurrence determination system according to claim 1,
Used by workers working at sites where power transmission and distribution facilities exist,
storing the analytical model;
and outputting information based on the objective variables obtained by inputting information based on sound data acquired at the site as an analysis target into the analysis model as the explanatory variables.
Worker terminal. A method for determining whether salt damage has occurred in power transmission and distribution equipment, comprising:
The information processing device
a step of storing an analytical model, which is a machine learning model trained using learning data configured using information based on sound data, which is environmental sound acquired at the site where the power transmission and distribution equipment is installed, as explanatory variables, and information indicating the presence or absence of leak sounds emitted from the power transmission and distribution equipment due to salt damage and information indicating the type of leak sounds if the leak sounds are present, as objective variables; and
a step of outputting information based on the objective variables obtained by inputting information based on sound data acquired at the site as an analysis target into the analysis model as the explanatory variables;
A method for determining whether salt damage has occurred. A method for determining whether salt damage has occurred in power transmission and distribution equipment, comprising:
The information processing device
a step of storing an analytical model, which is a machine learning model trained using learning data configured using information based on sound data, which is environmental sound acquired at the site where the power transmission and distribution equipment is installed, and measurement information, which is information acquired by an environmental sensor at the site, as explanatory variables, and information indicating the presence or absence of leak sounds emitted from the power transmission and distribution equipment due to salt damage and information indicating the type of leak sounds if the leak sounds are present, as objective variables; and
a step of outputting information based on the objective variables obtained by inputting information based on sound data and environmental measurement information acquired at the site as the analysis target into the analysis model as the explanatory variables;
A method for determining whether salt damage has occurred.