JP7757674B2 - Information processing device, information processing program, and information processing method - Google Patents

Description

The present invention relates to an information processing device, an information processing program, and an information processing method.

A method is known for using a prediction model to predict the amount of water flowing into a dam from the upstream area per unit time (dam inflow). Because the characteristics of the inflow into a dam change depending on seasonal factors, several prediction models have been developed to suit seasonal characteristics.

For example, Patent Document 1 discloses a prediction model that is generated taking into account rainfall and resulting snowmelt when there is snowfall upstream of a dam.

Patent Document 2 also discloses a prediction model that improves prediction accuracy by using rainfall intensity values obtained in each of multiple divided areas when rainfall occurs upstream of a dam.

JP 2008-185489 A Patent No. 6222689

When predicting the inflow volume into a dam, it is necessary to appropriately select one prediction model from the multiple prediction models described above; for example, a prediction model is automatically selected depending on the date of the prediction target.

However, climate change is subject to uncertainty; for example, the timing of snow accumulation and melting can vary from year to year. Therefore, the forecast model selected depending on the date of the forecast may not be appropriate.

The object of the present invention is to provide an information processing device that can automatically select an appropriate prediction model for predicting inflow into a dam.

One invention to achieve the above object is an information processing device that selects a prediction model for predicting inflow into a dam, the information processing device comprising: an acquisition unit that acquires weather information, including the presence or absence of snowfall, temperature, and precipitation, for a target prediction date at a prediction location; and a selection unit that selects a first prediction model based on snowfall and temperature as the prediction model for predicting inflow into the target dam when there is snowfall, the temperature is higher than a first temperature, and the precipitation is less than the first precipitation at the prediction location. Other features of the invention will become clearer in the description of this specification.

The present invention provides an information processing device that can automatically select an appropriate prediction model for predicting inflow into a dam.

FIG. 1 is a diagram illustrating an information processing system. FIG. 2 illustrates an example of a hardware configuration of an information processing device. FIG. 2 is a diagram illustrating an example of functional blocks realized in the information processing device. FIG. 10 is a diagram illustrating a method for estimating the amount of snowfall. FIG. 10 is a diagram illustrating the application range of each prediction model. FIG. 10 is a diagram illustrating the application range of each prediction model. 10 is a flowchart illustrating the flow of processing up to selection of a prediction model. FIG. 2 is a diagram illustrating information related to past weather.

==Embodiment==
<<Information Processing System>>
FIG. 1 is a diagram illustrating an overview of an information processing system 1 according to this embodiment. The information processing system 1 is a system for selecting an appropriate prediction model for predicting the amount of water inflow into a dam (described below) that is a target of prediction. The information processing system 1 includes a meteorological observation device 4, an information processing device 5, and a server 6. FIG. 1 also illustrates a river including an upstream dam 2, a downstream dam 3, a main river R1, and a tributary R2 that joins the main river R1. First, the upstream dam 2 and the downstream dam 3 will be described below.

<Upstream dam>
In this example, the upstream dam 2 is located upstream of the confluence of the main river R1 and the tributary R2. The upstream dam 2 is equipped with a water level gauge for measuring the level of the stored water, a gate for discharging the stored water, and a gate opening meter for measuring the gate opening (not shown).

<Downstream dam>
The downstream dam 3 is a dam that is the target of prediction in the information processing system 1. A prediction model that predicts the amount of water inflow into the downstream dam 3 is selected by the information processing system 1. In other words, the underlined dam corresponds to the "dam that is the target of prediction."

The downstream dam 3 is located downstream of the confluence of the main river R1 and tributary R2. Water flowing down the main river R1 and tributary R2 flows into the downstream dam 3. Here, the water flowing down the main river R1 includes water caused by meteorological factors such as rainfall, and water released from the upstream dam 2.

Like the upstream dam 2, the downstream dam 3 is equipped with a water level gauge, gate, and opening gauge (not shown).

<Weather observation equipment>
The meteorological observation device 4 is a device for observing the weather at a location that affects the inflow amount to the downstream dam 3. The meteorological observation device 4 is installed at a predetermined location upstream of the downstream dam 3. Hereinafter, the location where the meteorological observation device 4 is installed will be referred to as the "observation location."

The meteorological observation device 4 of this embodiment is a device that measures the amount of snow accumulation, temperature, and precipitation at an observation point, and is composed of a snow gauge, a temperature sensor, and a rain gauge. Here, "precipitation" includes rainfall or snowfall, and "precipitation amount" is the volume of rainfall or snowfall per unit area converted into water. Note that a meteorological observation device 4 may be provided at each of multiple observation points.

Hereinafter, information including the amount of snowfall, temperature, and precipitation will be referred to as "weather-related information." Note that in this embodiment, weather-related information is information including the amount of snowfall, temperature, and precipitation, but is not limited to these and may also include, for example, wind direction, humidity, etc.

Instead of the meteorological observation device 4, a snow depth gauge, a temperature sensor, and a rain gauge may each be installed individually at each observation point. If there are multiple observation points, at least one of a snow depth gauge, a temperature sensor, and a rain gauge may be installed at each of the multiple observation points. In other words, it is sufficient that at least one of each of snow depth, temperature, and precipitation is observed in an observation area that includes one or multiple observation points.

In this case, it is preferable that the number of observation points equipped with rain gauges be the largest in order to improve the accuracy of meteorological information and more accurately predict the inflow volume into the target dam. After the number of observation points equipped with rain gauges, it is also preferable that the number of observation points equipped with temperature sensors be the second largest. The meteorological observation device 4 transmits the observed meteorological information to the server 6.

<Server>
The server 6 receives information about the weather observed by the weather observation device 4 from the weather observation device 4 and stores it in an auxiliary storage device (not shown). Therefore, the server 6 stores information about past weather at the observation point (amount of snowfall, temperature, amount of precipitation).

<Information processing device>
The information processing device 5 is a device for selecting a prediction model for predicting the inflow amount to the downstream dam 3 on the target prediction date based on weather forecasts and various data acquired from the meteorological observation device 4. The hardware configuration of the information processing device 5 and the realized functional blocks will be described below.

Hardware Configuration FIG. 2 is a diagram illustrating an example of hardware of the information processing device 5. The illustrated information processing device 5 includes a processor 500, a main memory device 501, an auxiliary memory device 502, an input device 503, an output device 504, and a communication device 505. Note that all or part of the illustrated information processing device 5 may be realized using virtual information processing resources provided using virtualization technology, process space separation technology, or the like, such as a virtual server provided by a cloud system. Furthermore, all or part of the functions provided by the information processing device 5 may be realized by a service provided by the cloud system via an API (Application Programming Interface), for example.

The processor 500 is configured using, for example, a CPU (Central Processing Unit) etc.

The main memory device 501 is a device that stores programs and data, and is, for example, a ROM (Read Only Memory) or a RAM (Random Access Memory).

The auxiliary storage device 502 is, for example, a hard disk drive, a storage area of a cloud server, etc. Programs and data can be read into the auxiliary storage device 502 via a recording medium reader or a communication device 505. Programs and data stored in the auxiliary storage device 502 are read into the main storage device 501 as needed.

The input device 503 is an interface that accepts input from outside, such as a keyboard, mouse, or touch panel.

The output device 504 is an interface that outputs various information such as the progress and results of processing. The output device 504 is, for example, a display device (such as an LCD monitor) that visualizes the various information described above, a device that converts the various information described above into audio (an audio output device (such as a speaker)), or a device that converts the various information described above into text (such as a printer).

The input device 503 and output device 504 constitute a user interface that accepts and presents information to the user.

The communication device 505 is a device that enables communication with other devices. The communication device 505 is a wired or wireless communication interface that enables communication with other devices via a communication network NW such as the Internet.

The information processing device 5 may be equipped with, for example, an operating system, a file system, etc.

Each function of the information processing system 1 is realized by the processor 500 of the information processing device 5 reading and executing a program stored in the main memory device 501, or by the hardware that makes up the information processing system 1. The information processing system 1 stores various types of information (data), for example, as database tables or files managed by a file system.

- Use of information processing device 5 As described above, the information processing device 5 of this embodiment selects a prediction model for predicting the inflow volume to the downstream dam 3 on the target prediction date based on weather forecasts, information on past weather, etc., and predicts the inflow volume based on the selected prediction model.

Here, in order to predict the inflow volume into the downstream dam 3, it is necessary to know information about the weather on the prediction day at a point that will affect the inflow volume (hereinafter referred to as the "prediction point"). A prediction point is a point that has been pre-selected as a point where the weather information at that point will affect the inflow volume into the downstream dam 3. Specifically, a specified point upstream of the downstream dam 3 is selected as the prediction point.

Of the weather information for the target day, temperature and precipitation can generally be obtained from weather forecasts. However, snowfall amount cannot be obtained from weather forecasts, or even if it can be obtained from weather forecasts, the prediction accuracy may be insufficient. Therefore, in such cases, it is necessary to estimate the amount of snowfall at the forecast location on the target day from the past amount of snowfall at the observation location.

Furthermore, to accurately predict inflow volume, it is necessary to select an appropriate prediction model from multiple different prediction models depending on weather conditions. In this embodiment, five types of prediction models, M1 to M5, are created, and these prediction models can be selected when predicting flow rate. Each of the prediction models M1 to M5 will be explained below.

・Prediction model M1 (snow accumulation and temperature)
The prediction model M1 is a prediction model based on snow accumulation and temperature. Specifically, the prediction model M1 takes into account the effect of snow accumulation at the prediction point melting due to the influence of temperature, turning into water, and flowing into the downstream dam 3. The prediction model M1 inputs the amount of snow accumulation, temperature, and precipitation, among other meteorological information, and outputs the inflow amount.

・Prediction model M2 (snow accumulation, temperature and precipitation)
The prediction model M2 is a prediction model based on snow accumulation, temperature, and precipitation. Specifically, the prediction model M2 takes into account the effect of snow accumulation at the prediction point melting due to the influence of temperature and precipitation, turning into water, and flowing into the downstream dam 3, as well as the effect of precipitation at the prediction point flowing directly into the downstream dam 3. The prediction model M2 inputs the amount of snow accumulation, temperature, and precipitation, among other meteorological information, and outputs the inflow amount.

・Prediction model M3 (snow and precipitation)
The prediction model M3 is a prediction model based on snow accumulation and precipitation. Specifically, the prediction model M3 is a prediction model that takes into account the effect of precipitation (rainfall or snowfall) at the prediction point flowing into the downstream dam 3. The prediction model M3 takes the amount of snow accumulation and the amount of precipitation, which are information related to the weather, as input, and outputs the amount of inflow.

・Prediction model M4 (rainfall)
The prediction model M4 is a rainfall-based prediction model. Specifically, the prediction model M4 is a prediction model that takes into account the effect of precipitation (rainfall) at the prediction point flowing into the downstream dam 3. The prediction model M4 receives inputs of temperature and precipitation amount from meteorological information, and outputs the inflow amount.

・Prediction model M5 (snowfall)
The prediction model M5 is a prediction model based on snowfall. Specifically, the prediction model M5 is a prediction model that takes into account the effect of precipitation (snowfall) at the prediction point flowing into the downstream dam 3. The prediction model M5 takes temperature and precipitation amount, which are information related to the weather, as input and outputs the inflow amount.

As will be described in more detail below, at least one of these prediction models M1 to M5 can be a model created using, for example, a regression equation or a neural network. As will be described in more detail below, prediction models M1 to M5 are created using information about past weather at the prediction location and the corresponding past inflows into the dam at the prediction location.

3 is a diagram showing an example of functional blocks realized in the information processing device 5. When the processor 500 of the information processing device 5 executes a predetermined program, an acquisition unit 510, a model creation unit 511, a snow accumulation amount estimation unit 512, a selection unit 513, and an inflow amount prediction unit 514 are realized in the information processing device 5.

[Acquisition part]
The acquisition unit 510 acquires creation information Im for creating a prediction model and selection information Is for selecting a prediction model. Details of each piece of information and the process by which the acquisition unit 510 acquires each piece of information will be described below.

Creation information Im
The creation information Im is information about the past weather at the observation point, which is used when the model creation unit 511 (described later) creates a prediction model. Details will be explained together with the explanation of the model creation unit 511, but first, information about the past weather at the prediction point is estimated based on information about the past weather at the observation point. Then, a prediction model is created based on the information about the past weather at the prediction point.

The acquisition unit 510 acquires information about past weather at observation points recorded in an auxiliary storage device (e.g., a hard disk drive) (not shown) of the server 6 via a communication network NW, etc.

If information about the past weather at the prediction location is already stored in the auxiliary storage device of server 6, the information about the past weather at the prediction location can be used as creation information Im instead of the information about the past weather at the observation location. Even in such a case, a prediction model is created based on the information about the past weather at the prediction location.

Alternatively, the acquisition unit 510 may acquire past weather forecast values as information regarding the past weather at the prediction location. In this case, the acquisition unit 510 may acquire forecast data such as GPV data that was previously distributed regularly by the Japan Meteorological Agency, for example, via a communication network NW or the like.

Selection information Is
The forecast model used to forecast the flow rate of a target dam differs depending on the weather information (snowfall, temperature, precipitation) at the forecast location on the target forecast date. Therefore, to select an appropriate forecast model from forecast models M1 to M5, information about the weather at the forecast location is first required.

Here, among the information about the weather at the prediction location on the target prediction date (snow accumulation, temperature, precipitation), temperature and precipitation can be obtained, for example, from weather forecasts, but snow accumulation is generally difficult to obtain. However, the amount of snow accumulation at the prediction location on the target prediction date can be estimated, for example, based on the past trends in snow accumulation at the observation location.

For this reason, in this embodiment, the selection information Is for selecting a prediction model is the past trends in snowfall at the observation point, the estimated temperature and estimated precipitation for the target day at the prediction point.

Details of Acquisition Unit 510 The acquisition unit 510 acquires, from the selection information Is, the change in the amount of snowfall at the observation point in the past from an auxiliary storage device (not shown) of the server 6 via the communication network NW or the like.

Furthermore, the acquisition unit 510 acquires estimated values of the temperature and precipitation for the prediction date at the prediction location from the selection information Is via a weather forecast via a communication network NW, etc. Note that the weather forecast may use forecast data such as GPV data regularly distributed by the Japan Meteorological Agency, for example.

However, if all of the estimated values for the temperature, precipitation, and snowfall amount for the target prediction day at the prediction location are obtained from the weather forecast, these may be used as the selected information Is. In this case, the acquisition unit 510 will acquire the estimated values for the snowfall amount, temperature, and precipitation amount for the target prediction day at the prediction location from the weather forecast via a communication network NW, etc.

The selection information Is is not limited to estimated values of weather information for the target prediction day described above, but may also use weather information as current or past actual values. In this case, weather information as actual values is obtained from server 6. For example, if the target prediction day is the current day, weather information as current actual values may be used, or weather information from a past day that is similar to the weather conditions of the current day may be used.

[Model Creation Department]
The model creation unit 511 creates prediction models M1 to M5 that predict the inflow amount to the downstream dam 3. As described above, each of the prediction models M1 to M5 is a model that takes some or all of the weather information (snowfall amount, temperature, precipitation amount) at the prediction point as input and outputs the inflow amount to the downstream dam 3.

The model creation unit 511 creates prediction models M1 to M5 based on information about past weather at the prediction location and the corresponding past inflow volume into the downstream dam 3.

Information about the past weather at the prediction location can be estimated from information about the past weather at the observation location acquired by the acquisition unit 510 as creation information Im. For example, if the prediction location and the observation location are sufficiently close to each other, information about the weather at the prediction location may be considered to be equivalent to information about the weather at the observation location.

Alternatively, if multiple observation points are set up, information about the weather at the prediction point may be estimated based on information about the weather at each of the multiple observation points. For example, the average or weighted average of the precipitation amounts at each of the multiple observation points may be used as the precipitation amount at the prediction point.

Alternatively, information about past weather conditions at the prediction location may be obtained from forecast values of past weather forecasts.

At this time, the model creation unit 511 creates prediction models M1 to M5 using, for example, a regression equation or a neural network, so that when information about past weather at a prediction location is input, the corresponding past inflow amount to the dam at the prediction location is output.

Specifically, the model creation unit 511 generates a prediction model M1 based on past snowfall and temperature at the prediction location and past inflow volume, and generates a prediction model M2 based on past snowfall, temperature, and rainfall at the prediction location and past inflow volume.

The model creation unit 511 also generates a prediction model M3 based on past snowfall and precipitation at the prediction location and past inflow volume, and generates a prediction model M4 based on past rainfall and past inflow volume at the prediction location. The model creation unit 511 also generates a prediction model M5 based on past snowfall and past inflow volume at the prediction location.

[Snowfall estimation section]
The snowfall amount estimation unit 512 estimates the amount of snowfall at a predetermined prediction point on the prediction target day. The snowfall amount estimation unit 512 estimates the amount of snowfall based on at least the observation results of the meteorological observation device 4.

Note that the weather observation device 4 is a device that can measure the amount of snowfall in real time, but is not a device that can measure the amount of snowfall on the target forecast day. Also, as mentioned above, the weather observation device 4 is installed at an observation point. However, the observation point and the forecast point are not necessarily the same location.

The snow accumulation estimation unit 512 therefore estimates the amount of snow accumulation for the specified day based on the actual snow accumulation value for a specified day at an observation point distant from the prediction point, the prediction point, the observation point, the prediction date, and the specified day.

The method for estimating the amount of snowfall at a prediction location on a target day will be explained using Figure 4. Figure 4 is a diagram explaining the method for estimating the amount of snowfall, and is a graph with time on the horizontal axis and snowfall on the vertical axis.

P0, P1, and P2 in the figure represent a prediction point, an example of an observation point, and another example of an observation point, respectively. Here, the elevation of observation point P1 is higher than the elevation of prediction point P0, and the elevation of observation point P2 is lower than the elevation of prediction point P0. Figure 4 shows the changes in snow accumulation at prediction point P0, observation point P1, and observation point P2. Also, D0 in the figure indicates the prediction target date. D1 in the figure indicates a specified day in the past on which the snow accumulation at observation point P2 was zero.

In the example of Figure 4, the snow accumulation amounts at prediction point P0, observation point P1, and observation point P2 change while maintaining a roughly constant difference between them. In this example, the snow accumulation amounts at prediction point P0, observation point P1, and observation point P2 decrease monotonically.

In other words, the difference in snow accumulation between prediction point P0, observation point P1, and observation point P2 can be estimated from the difference in elevation between prediction point P0, observation point P1, and observation point P2.

From this, the difference between the dates and times when the snow accumulation amounts at prediction point P0, observation point P1, and observation point P2 are the same can be estimated from the elevation difference between prediction point P0, observation point P1, and observation point P2.

In the example of Figure 4, the difference in date and time when the snow accumulation amounts at prediction point P0 and observation point P2 are the same is shown as ΔD. In this case, the snow accumulation amount at observation point P2 on date D1 corresponds to the snow accumulation amount at prediction point P0 on date D1 plus ΔD. This makes it possible to estimate the change in snow accumulation amount at prediction point P0 based on the change in snow accumulation amount at observation point P2.

In this example, since D1 + ΔD < D0, it can be estimated that there will be no snowfall at prediction point P0 on prediction target day D0.

[Selection section]
The selection unit 513 selects an appropriate prediction model for predicting the inflow amount to the downstream dam 3 according to information about the weather at the prediction point on the target prediction day. When selecting a prediction model, the selection unit 513 makes the following multiple determinations based on the information for selecting a prediction model acquired by the acquisition unit 510.

The selection unit 513 determines whether or not there is snow at the prediction location. The selection unit 513 determines whether or not there is snow based on the amount of snow at the prediction location estimated by the snow amount estimation unit 512.

Here, "presence or absence of snow" means whether the amount of snow is 0 (no snow) or whether the amount of snow is greater than 0 (snow).

The selection unit 513 may set a threshold value for the maximum amount of snow accumulation at which the amount of snow accumulation can be considered to be essentially zero, and may consider the amount of snow accumulation to be zero (no snow accumulation) if the amount of snow accumulation estimated by the snow accumulation estimation unit 512 is equal to or less than the threshold value.

The selection unit 513 also determines whether the air temperature at the prediction location is higher than air temperature T1 (corresponding to the first air temperature).

If the temperature at the forecast location is higher than T1, it means that if there is precipitation at the forecast location, there will be snowmelt due to temperature. On the other hand, if the temperature at the forecast location is below T1, it means that if there is precipitation at the forecast location, there will be no snowmelt due to temperature.

The user should set T1 in advance, taking the above into consideration. Note that this temperature T1 determination can be made only when it is determined that there is snow accumulation at the prediction location.

The selection unit 513 also determines whether the amount of precipitation at the prediction location is less than W1 (corresponding to the first amount of precipitation). W1 is the amount of precipitation preset by the user. This determination can be made if it is determined that there is snowfall at the prediction location and the temperature is higher than T1.

The selection unit 513 also determines whether the air temperature at the prediction location is higher than a predetermined air temperature T2 (corresponding to a second air temperature).

If the temperature at the forecast location is higher than T2, it means that if there is precipitation at the forecast location, it is predicted to be rain. On the other hand, if the temperature at the forecast location is below T2, it means that if there is precipitation at the forecast location, it is predicted to be snow.

The user should set T2 in advance, taking the above into consideration. This determination can be made only when it is determined that there is no snow accumulation at the prediction location.

The selection unit 513 selects an appropriate prediction model for predicting the inflow volume into the downstream dam 3 based on a determination based on information about the weather at the prediction location on the target prediction day.

The selection process performed by the selection unit 513 will be explained using Figures 5 and 6. Figures 5 and 6 show a two-dimensional space with the horizontal axis representing the temperature at the prediction point and the vertical axis representing the amount of precipitation at the prediction point. Figure 5 shows the case where there is snow at the prediction point, and Figure 6 shows the case where there is no snow at the prediction point.

Figure 5 shows regions 1 to 3 in two-dimensional space, with their respective boundaries indicated by solid lines. Region 1 is a region where the temperature is higher than T1 and the amount of precipitation is less than W1. Region 2 is a region where the temperature is higher than T1 and the amount of precipitation is equal to or greater than W1. Region 3 is a region where the temperature is equal to or less than T1.

Figure 6 shows regions 4 and 5 in two-dimensional space, with their respective boundaries indicated by solid lines. Region 4 is the region where the temperature is higher than T2. Region 5 is the region where the temperature is lower than T2.

The selection unit 513 selects an appropriate prediction model for predicting the inflow volume to the downstream dam 3 based on the corresponding area in Figure 5 or 6 for the weather information for the prediction date at the prediction point.

In this embodiment, the selection unit 513 selects one prediction model from five prediction models M1 to M5. Below, we will explain how the selection unit 513 selects each of the prediction models M1 to M5.

When prediction model M1 is selected When there is snow accumulation at the prediction point, the temperature is higher than temperature T1, and the amount of precipitation is less than precipitation W1, the selection unit 513 selects prediction model M1 (corresponding to the first prediction model) as the prediction model for predicting the inflow amount to the downstream dam 3. When the temperature at the prediction point is higher than T1, it means that there is snowmelt due to temperature at the prediction point. This case corresponds to the case where the information on the weather at the prediction point is within region 1 in Figure 5.

When prediction model M2 is selected: If there is snowfall at the prediction point, the temperature is higher than temperature T1, and the precipitation is equal to or greater than precipitation W1, the selector 513 selects prediction model M2 (corresponding to the second prediction model) as the prediction model. This case corresponds to the case where the weather information at the prediction point is within region 2 in FIG. 5 .

When prediction model M3 is selected When there is snowfall and the temperature is equal to or lower than temperature T1 at the prediction point, the selection unit 513 selects prediction model M3 (corresponding to the third prediction model) as the prediction model. The temperature being equal to or lower than T1 at the prediction point means that there is no snowmelt due to temperature at the prediction point. This case corresponds to the case where the weather information at the prediction point is within region 3 in FIG. 5.

When prediction model M4 is selected: If there is no snowfall and the temperature is higher than temperature T2 at the prediction point, the selection unit 513 selects rainfall-based prediction model M4 (corresponding to the fourth prediction model) as the prediction model. If the temperature at the prediction point is higher than T2, it means that if there is precipitation at the prediction point, it will be rainfall. This case corresponds to the case where the information about the weather at the prediction point is within region 4 in Figure 6.

When prediction model M5 is selected: When there is no snowfall and the temperature at the prediction point is equal to or lower than temperature T2, the selection unit 513 selects prediction model M5 (fifth prediction model) as the prediction model. When the temperature at the prediction point is equal to or lower than T2, it means that if there is precipitation at the prediction point, it will be snowfall. This case corresponds to the case where the information about the weather at the prediction point is within region 5 in FIG. 6.

[Inflow volume prediction section]
The inflow prediction unit 514 predicts the inflow to the downstream dam 3 on the target prediction day based on information about the weather on the target prediction day and the selected prediction model. The information about the weather on the target prediction day is the amount of snowfall, temperature, and precipitation estimated from the selected information Is acquired by the aforementioned acquisition unit 510. By inputting this information about the weather into the selected prediction model, the inflow to the downstream dam 3 on the target prediction day is output.

With the above configuration, the information processing device 5 can select an appropriate prediction model and predict the inflow volume to the downstream dam 3 on the target prediction day.

<Processing up to selecting a prediction model>
7 is a flowchart illustrating the flow of processing up to the information processing device 5 selecting a prediction model. Note that the following description will be given assuming that the prediction models M1 to M5 to be selected have already been created by the model creation unit 511 based on the creation information Im acquired by the acquisition unit 510. The processing up to the selection of a prediction model includes steps S101 to S111.

First, in step S101, the acquisition unit 510 acquires the selection information Is. Next, in step S102, the snow accumulation estimation unit 512 estimates the snow accumulation amount for the prediction date at the prediction location.

Next, in step S103, the selection unit 513 determines whether or not there is snow at the prediction location based on the amount of snow estimated by the snow amount estimation unit 512 in step S102.

The presence or absence of snow accumulation means whether the amount of snow accumulation is 0 (no snow accumulation) or whether the amount of snow accumulation is greater than 0 (snow accumulation). It is also possible to set a threshold value for the maximum amount of snow accumulation at which the amount of snow accumulation can be considered to be essentially 0, and to consider the amount of snow accumulation to be 0 (no snow accumulation) if the amount of snow accumulation estimated by the snow accumulation estimation unit 512 in step S102 is equal to or less than the threshold value.

If the selection unit 513 determines in step S103 that there is snow (step S103: Y), the process proceeds to step S104, where the selection unit 513 determines whether the air temperature is higher than air temperature T1.

If the selection unit 513 determines in step S104 that the temperature is higher than T1 (step S104: Y), the process proceeds to step S105, where the selection unit 513 determines whether the precipitation amount is less than precipitation amount W1.

If, in step S105, the selection unit 513 determines that the amount of precipitation is less than W1 (step S105: Y), the process proceeds to step S106, where the selection unit 513 selects the prediction model M1.

On the other hand, if in step 105 the selection unit 513 determines that the precipitation amount is equal to or greater than W1 (step 105: N), the process proceeds to step 107, where the selection unit 513 selects prediction model M2.

Returning to step 104, if the selection unit 513 determines that the temperature is below T1 (step 104: N), the process proceeds to step 108, where the selection unit 513 selects prediction model M3.

Returning to step 103, if the selection unit 513 determines that there is no snow (step 103: N), the process proceeds to step 109, where the selection unit determines whether the air temperature is higher than air temperature T2.

If, in step 109, the selection unit 513 determines that the temperature is higher than T2 (step 109: Y), the process proceeds to step 110, where the selection unit 513 selects prediction model M4.

On the other hand, if in step 109 the selection unit 513 determines that the temperature is equal to or lower than T2 (step 109: N), the process proceeds to step 111, where the selection unit 513 selects prediction model M5.

Through the above processing, the information processing device 5 can select an appropriate prediction model.

==Variation 1==
In the above embodiment, the prediction models M1 to M5 are exemplified by regression equations or neural networks, but are not limited to these. In particular, the prediction model M1 may be a prediction model generated based on the similarity between information about the weather on the target day and information about past weather, and past inflows into the dam. One example of such a prediction model is so-called Just-In-Time (JIT) modeling, which will be described in detail below.

Figure 8 is a diagram showing an example of information 7 related to past weather. Information 7 related to past weather is information related to weather over a certain period of time in the past. In this example, the information includes information related to weather (snowfall, temperature, and precipitation) for each hour of each day over the one year period from January 1, 2020 to December 31, 2020, as well as data on the amount of inflow into the downstream dam 3.

In this modified example, the similarity can be, for example, cosine similarity. Specifically, a three-dimensional vector having components related to weather, i.e., normalized values of snowfall, temperature, and precipitation, is generated for the target prediction date and each past day. The dot product of the three-dimensional vectors for the target prediction date and each past day corresponds to the cosine similarity.

Then, from the past days for one year shown in Figure 8, a predetermined number of past days (one or more) are extracted in descending order of similarity. The average value of the inflow volume into the dam for the extracted past days is used as the predicted inflow volume for the target prediction day.

By using such JIT modeling, the prediction accuracy of the prediction model M1 in particular can be further improved. Note that JIT modeling may be used not only for the prediction model M1 but also for the prediction models M2 to M4.
==Variation 2==
In the above embodiment, the prediction point and the observation point are distant from each other. However, the prediction point and the observation point may be the same point. In other words, the meteorological observation device 4 may be installed at the prediction point.

In this case, the selection unit 513 may determine whether there will be snow at the prediction location on the target prediction date based on the actual snowfall amount at the prediction location on the target prediction date, and the target prediction date.

As an example, if there is snow at the prediction point on a specific day, the day on which the snow amount will be zero can be predicted based on the change in the amount of snow at the prediction point. If the prediction target day is before the day on which the snow amount is predicted to be zero, the selection unit 513 can determine that there will be snow on the prediction target day. If the prediction target day is after the day on which the snow amount is predicted to be zero, the selection unit 513 can determine that there will be no snow on the prediction target day.

As another example, if there is snow at the prediction location on a specific day, the amount of snow at a future date and time can be estimated based on the change in the amount of snow at the prediction location. This makes it possible to estimate the amount of snow on the target prediction date and ultimately determine whether there will be snow.

==Summary==

As described above, in the information processing device 5 of the embodiment that selects a prediction model for predicting the inflow amount to a dam, the information processing device 5 includes an acquisition unit 510 that acquires information related to the weather, including the amount of snowfall, temperature, and precipitation, for the target prediction date at the prediction location, and a selection unit 513 that selects a first prediction model based on snowfall and temperature as the prediction model for predicting the inflow amount to the target dam when there is snowfall, the temperature is higher than the first temperature, and the precipitation is less than the first precipitation at the prediction location.

With this configuration, when it is appropriate to use the first forecasting model based on snow depth and temperature, the first forecasting model can be automatically selected.

In the information processing device 5, when there is snow accumulation, the temperature is higher than the first temperature, and the precipitation is equal to or greater than the first precipitation at the prediction location, the selection unit 513 selects the second prediction model based on snow accumulation, temperature, and precipitation as the prediction model.

With this configuration, when it is appropriate to use a second forecasting model based on snowfall, temperature, and precipitation, the second forecasting model can be automatically selected.

In the information processing device 5, the selection unit 513 selects the third prediction model based on snow accumulation and precipitation as the prediction model when there is snow accumulation and the temperature is equal to or lower than the first temperature at the prediction location. With this configuration, the third prediction model based on snow accumulation and precipitation can be automatically selected when it is appropriate to use the third prediction model.

In the information processing device 5, the selection unit 513 determines whether the air temperature at the prediction location is higher than the second air temperature, and if there is no snowfall and the air temperature at the prediction location is higher than the second air temperature, the selection unit 513 selects the fourth prediction model based on rainfall as the prediction model. With this configuration, the fourth prediction model based on rainfall can be automatically selected when it is appropriate to use the fourth prediction model.

In the information processing device 5, the selection unit 513 selects the fifth prediction model based on snowfall as the prediction model when there is no snowfall and the temperature is equal to or lower than the second temperature at the prediction location. With this configuration, the fifth prediction model based on snowfall can be automatically selected when it is appropriate to use the fifth prediction model.

In the information processing device 5, the first prediction model is a prediction model generated based on the degree of similarity between information about the weather on the prediction date and information about past weather, as well as past inflows into the dam. This configuration further improves the prediction accuracy of the first prediction model.

In the information processing device 5, at least one of the prediction models is a model generated using a regression equation or a neural network. This configuration improves the prediction accuracy of the prediction model used.

The information processing device 5 further includes a snow accumulation estimation unit 512 that estimates the amount of snow accumulation at the prediction point on a specified day based on the actual snow accumulation value for an observation point distant from the prediction point, the prediction point, the observation point, the target prediction day, and the specified day. With this configuration, it is possible to estimate the amount of snow accumulation at the prediction point even if a snow accumulation meter is not installed at the prediction point where the amount of snow accumulation is to be estimated. This improves the validity of the selected prediction model.

The information processing device 5 may further include a snow accumulation estimation unit 512 that estimates the amount of snow accumulation at the prediction location on a specified day based on the actual snow accumulation value at the prediction location on that day, the target prediction day, and the specified day. This configuration makes it possible to determine whether or not there is snow accumulation at the prediction location. This improves the validity of the selected prediction model.

The information processing device 5 further includes an inflow prediction unit 514 that predicts the inflow into the dam on the target prediction day based on meteorological information and a selected prediction model. With this configuration, the inflow into the dam can be predicted using an appropriate prediction model, thereby improving prediction accuracy.

In addition, the information processing program of the embodiment is an information processing program for selecting a prediction model for predicting the inflow amount into a dam, and includes an acquisition unit 510 that acquires weather information including the amount of snowfall, temperature, and precipitation for the target prediction date at the prediction location, and a selection unit 513 that selects a first prediction model based on snowfall and temperature as the prediction model for predicting the inflow amount into the target dam when there is snowfall, the temperature is higher than the first temperature, and the precipitation is less than the first precipitation at the prediction location.

Another embodiment of an information processing method is an information processing method for selecting a prediction model for predicting the inflow volume into a dam, and includes the steps of acquiring weather information including the amount of snowfall, temperature, and precipitation for the target prediction date at the prediction location, and selecting a first prediction model based on snowfall and temperature as the prediction model for predicting the inflow volume into the target dam when there is snowfall, the temperature is higher than a first temperature, and the precipitation is less than the first precipitation at the prediction location.

The above embodiments are intended to facilitate understanding of the present invention and are not intended to limit the scope of the present invention. Furthermore, the present invention may be modified or improved without departing from its spirit, and it goes without saying that the present invention includes equivalents thereof.

For example, in the above embodiment, the information processing device 5 is shown to have an inflow volume prediction unit 514 that predicts inflow volumes at future times using the generated prediction model. However, the inflow volume prediction unit 514 may not be provided in the information processing device 5, but may be provided in another device.

1: Information processing system 2: Upstream dam 3: Downstream dam 4: Meteorological observation device 5: Information processing device 6: Server 7: Information related to past weather 500: Processor 501: Main memory device 502: Auxiliary memory device 503: Input device 504: Output device 505: Communication device 510: Acquisition unit 511: Model creation unit 512: Snow accumulation amount estimation unit 513: Selection unit 514: Inflow amount prediction unit

Claims (12)

An information processing device that selects a prediction model for predicting an inflow amount into a dam, the prediction model being created so that when information about past weather at a prediction point is input, a corresponding past inflow amount into the dam at the prediction point is output ,
an acquisition unit that acquires information about weather, including snowfall, temperature, and precipitation, for a target day at a prediction location;
a selection unit that selects, when there is snowfall, the temperature is higher than a first temperature, and the precipitation is less than a first precipitation at the prediction point, a first prediction model based on information about the weather including at least snowfall and temperature and past inflows into the dam as the prediction model for predicting the inflow into the dam to be predicted;
Equipped with
The first prediction model is
A prediction model that takes into account the effect of snow melting at the prediction point due to the influence of temperature, turning into water, and flowing into the dam,
The first temperature is
The temperature is used to identify whether or not there is snowmelt due to temperature when there is precipitation at the prediction point.
Information processing device. 2. The information processing device according to claim 1,
The selection unit
When there is snowfall, the temperature is higher than the first temperature, and the precipitation is equal to or greater than the first precipitation at the prediction point, a second prediction model is selected as the prediction model based on information about the weather including at least snowfall, temperature, and precipitation, and on the past inflow amount into the dam ;
The second prediction model is
The prediction model takes into account the effect of snow at the prediction point melting due to the influence of temperature and precipitation, turning into water and flowing into the dam, and the effect of precipitation at the prediction point flowing directly into the dam.
Information processing device. 3. The information processing device according to claim 1,
The selection unit
When there is snowfall and the temperature is equal to or lower than the first temperature at the prediction point, a third prediction model based on information about the weather including at least snowfall and precipitation and a past inflow amount into the dam is selected as the prediction model;
The third prediction model is
A prediction model that takes into account the effect of rainfall or snowfall at the prediction point flowing into the dam,
Information processing device. The information processing device according to any one of claims 1 to 3,
The selection unit
When there is no snowfall and the temperature is higher than a second temperature at the prediction point, a fourth prediction model based on information about the weather including at least rainfall and a past inflow amount into the dam is selected as the prediction model;
The fourth prediction model is
A prediction model that takes into account the effect of rainfall at the prediction point flowing into the dam,
The second temperature is
An information processing device that is a temperature for identifying whether it is rain or snow if precipitation occurs at the predicted location . 5. The information processing device according to claim 4,
The selection unit
When there is no snowfall and the temperature is equal to or lower than the second temperature at the prediction point, a fifth prediction model based on information about the weather including at least snowfall and the past inflow amount into the dam is selected as the prediction model;
The fifth prediction model is
A prediction model that takes into account the effect of snowfall at the prediction point flowing into the dam,
Information processing device. The information processing device according to any one of claims 1 to 5,
The first prediction model is
a prediction model generated based on the similarity between information about the weather on the prediction target day and information about past weather, and the amount of inflow into the dam in the past ;
a prediction model that predicts the inflow amount to the dam based on the inflow amount to the dam for one or more predetermined number of past days extracted in order from a certain past period, starting from days with the greatest similarity to the prediction target day;
Information processing device. The information processing device according to any one of claims 1 to 6,
At least one of the prediction models is a model generated by a regression equation or a neural network;
The regression equation is created so that, when information about the weather at the prediction point is used as an explanatory variable, the inflow amount into the corresponding dam at the prediction point is used as a target variable, and information about the weather in the past at the prediction point is used as an input, the corresponding inflow amount into the dam in the past at the prediction point is output;
The neural network is trained so that, when information about the weather at the prediction point is input and the inflow amount into the corresponding dam at the prediction point is output, and when information about the weather in the past at the prediction point is input, the corresponding inflow amount into the dam in the past at the prediction point is output.
Information processing device. The information processing device according to any one of claims 1 to 7,
a snowfall amount estimation unit that estimates the snowfall amount at the prediction point on the target prediction day based on a change in snowfall amount at an observation point distant from the prediction point, the prediction point, the observation point, and the target prediction day ;
The snow accumulation amount estimation unit
estimating a difference in snowfall amount between the observation point and the prediction point based on an elevation difference between the observation point and the prediction point;
based on the difference in snowfall amount, estimating the difference in date and time when the snowfall amount at the observation point and the prediction point will be the same;
estimating the amount of snowfall at the prediction point on the prediction target day based on the change in the amount of snowfall at the observation point and the difference in date and time;
Information processing device. The information processing device according to any one of claims 1 to 7,
The method further includes a snow accumulation amount estimation unit that estimates the snow accumulation amount at the prediction point on the prediction target day based on the transition of the snow accumulation amount at the prediction point and the prediction target day.
Information processing device. The information processing device according to any one of claims 1 to 9,
an inflow prediction unit that predicts the inflow into the dam on the target prediction day based on the information related to the meteorological conditions and the selected prediction model. An information processing program for selecting a prediction model for predicting an inflow amount into a dam, the prediction model being created so that, when information about past weather at a prediction point is input, a corresponding past inflow amount into the dam at the prediction point is output ,
On the computer,
an acquisition unit that acquires information about weather, including snowfall, temperature, and precipitation, for a target day at a prediction location;
a selection unit that selects, when there is snowfall, the temperature is higher than a first temperature, and the precipitation is less than a first precipitation at the prediction point, a first prediction model based on information about the weather including at least snowfall and temperature and past inflows into the dam as the prediction model for predicting the inflow into the dam to be predicted;
To achieve this ,
The first prediction model is
A prediction model that takes into account the effect of snow melting at the prediction point due to the influence of temperature, turning into water, and flowing into the dam,
The first temperature is
The temperature is used to identify whether or not there is snowmelt due to temperature when there is precipitation at the prediction point.
Information processing program. 1. An information processing method for selecting a prediction model for predicting an inflow amount into a dam, the prediction model being created so that, when information about past weather at a prediction point is input, a corresponding past inflow amount into the dam at the prediction point is output , the method comprising:
acquiring information about weather including snowfall, temperature, and precipitation for a target day at a prediction location;
When there is snowfall, the temperature is higher than a first temperature, and the precipitation is less than a first precipitation at the prediction point, selecting a first prediction model based on information about the weather including at least snowfall and temperature and past inflows into the dam as the prediction model for predicting the inflow into the dam to be predicted;
Including,
The first prediction model is
A prediction model that takes into account the effect of snow melting at the prediction point due to the influence of temperature, turning into water, and flowing into the dam,
The first temperature is
The temperature is used to identify whether or not there is snowmelt due to temperature when there is precipitation at the prediction point.
Information processing methods.