Water demand forecasting program, information processing device, and water demand forecasting method

Abstract

The present invention provides an information processing device that enables accurate prediction of the amount of purified water that needs to be supplied to households and other locations. [Solution] A first learning model is generated by performing machine learning on multiple first training data sets, each corresponding to a plurality of first time zones, which each include first weather information indicating the weather conditions in the first region for each time zone, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time zones corresponding to each time zone. The generated first learning model is then stored in a memory unit.

Description

[Technical Field]

[0001] This disclosure relates to a water demand forecasting program, an information processing device, and a water demand forecasting method. [Background technology]

[0002] Water treatment plants are equipped with water treatment facilities that perform various operations on raw water (hereinafter also referred to as treated water), such as river water or well water. Specifically, such water treatment facilities produce purified water (hereinafter also referred to as treated water) that needs to be supplied to households, etc. (hereinafter simply referred to as households, etc.) in the target area by performing various operations on the treated water (see Patent Document 1). [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2006-233764 [Overview of the project] [Problems that the invention aims to solve]

[0004] In water treatment facilities like the one described above, it is desirable to accurately predict the amount of purified water that needs to be supplied to households, for example, or in other words, the amount of purified water that needs to be generated by the water treatment facility. [Means for solving the problem]

[0005] The water demand forecasting program in this disclosure generates a first learning model by performing machine learning on a plurality of first training data, each of which corresponds to a plurality of first time periods, and each first training data includes first weather information indicating the weather conditions in a first region for each time period, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time periods corresponding to each time period, and the computer executes the process of storing the generated first learning model in a storage unit. [Effects of the Invention]

[0006] According to the water demand forecasting program, information processing device, and water demand forecasting method described in this disclosure, it becomes possible to accurately predict the amount of purified water that needs to be supplied to households and the like. [Brief explanation of the drawing]

[0007] [Figure 1] Figure 1 is a diagram illustrating the configuration of the water treatment system 1000 in the first embodiment. [Figure 2] Figure 2 is a diagram illustrating the configuration of the water treatment equipment 100 in the first embodiment. [Figure 3] Figure 3 is a diagram illustrating the hardware configuration of the information processing device 1. [Figure 4] Figure 4 is a block diagram of the functions of the information processing device 1 in the first embodiment. [Figure 5] Figure 5 illustrates a specific example of the model learning process in the first embodiment. [Figure 6] Figure 6 illustrates a specific example of the model learning process in the first embodiment. [Figure 7] Figure 7 illustrates a specific example of the water demand forecasting process in the first embodiment. [Figure 8] Figure 8 illustrates a specific example of the water demand forecasting process in the first embodiment. [Figure 9] Figure 9 is a flowchart illustrating the model learning process in the first embodiment. [Figure 10] Figure 10 is a diagram illustrating a specific example of weather information DT1. [Figure 11] Figure 11 illustrates a specific example of water volume information DT2. [Figure 12] Figure 12 illustrates a specific example of the training data DT11. [Figure 13]FIG. 13 is a diagram for explaining a specific example of the teacher data DT12. [Figure 14] FIG. 14 is a flowchart diagram for explaining the water demand prediction process in the first embodiment. [Figure 15] FIG. 15 is a diagram for explaining a specific example of the weather information DT1a. [Figure 16] FIG. 16 is a diagram for explaining a specific example of the time-series data of the water volume information DT2a. [Figure 17] FIG. 17 is a diagram for explaining a specific example of the time information DT0. [Figure 18] FIG. 18 is a diagram for explaining a specific example of the time-series data of the water volume information DT2b.

Embodiments for Carrying Out the Invention

[0008] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. However, such description should not be construed in a limiting sense and does not limit the subject matter recited in the claims. Also, various changes, substitutions, and modifications can be made without departing from the spirit and scope of the present disclosure. Further, different embodiments can be appropriately combined.

[0009] [Water Treatment System 1000 in the First Embodiment] First, a configuration example of the water treatment system 1000 in the first embodiment will be described. FIG. 1 is a diagram for explaining the configuration of the water treatment system 1000 in the first embodiment. Also, FIG. 2 is a diagram for explaining the configuration of the water treatment facility 100 in the first embodiment. Note that the positions and numbers of pumps and pipes in the following examples are for illustration only and are not limited thereto.

[0010] The water treatment system 1000 is, for example, a water treatment system installed in a water purification plant. Specifically, the water treatment system 1000 includes, for example, an information processing device 1, a control device 2, and water treatment equipment 100. In the example shown in Figure 1, the information processing device 1 and the control device 2 can access each other via a network (not shown), such as Ethernet®.

[0011] The water treatment equipment 100 is, for example, equipment that produces treated water by performing water purification treatment on the water to be treated.

[0012] Specifically, as shown in Figure 2, the water treatment facility 100 includes, for example, a grit tank 11, an intake well 12, a mixing tank 13, a flocculation tank 14 (hereinafter also simply referred to as the flocculation tank 14), a sedimentation tank 15, a filtration tank 16, a water purification tank 17, a distribution tank 18, pumps P1, P2, P3, and a storage tank T.

[0013] The sedimentation basin 11 is a tank into which the water to be treated, for example, taken from a river, first flows, and is a tank that settles and removes sediment and other materials contained in the water to be treated.

[0014] Pump P1 is, for example, a pump installed in piping connecting the sedimentation basin 11 and the intake well 12. Specifically, pump P1 supplies, for example, the water to be treated from the sedimentation basin 11 to the intake well 12.

[0015] The intake well 12 is, for example, a tank that adjusts the amount of water to be treated supplied from the sedimentation basin 11 and supplies it to the mixing basin 13.

[0016] The mixing tank 13 is, for example, a tank into which a coagulant is injected into the water to be treated supplied from the intake well 12.

[0017] The floc formation tank 14 is a tank that, for example, stirs the water to be treated supplied from the mixing tank 13, thereby coagulating the suspended solids contained in the water to be treated supplied from the mixing tank 13 with a coagulant to form flocs.

[0018] The sedimentation tank 15 is, for example, a tank that settles and separates flocs contained in the water to be treated supplied from the floc formation tank 14 from the water to be treated.

[0019] The filtration tank 16 is a tank that filters the water to be treated supplied from the sedimentation tank 15 by using a filter body (not shown) made of, for example, sand or gravel.

[0020] The water purification reservoir 17 is a tank that temporarily stores the water to be treated supplied from the filtration tank 16 (for example, the water to be treated after chlorine disinfection has been performed downstream of the filtration tank 16) and supplies it to the distribution reservoir 18.

[0021] Pump P2 is, for example, a pump installed in a pipe connecting the water purification reservoir 17 and the water distribution reservoir 18. Specifically, pump P2 supplies treated water from the water purification reservoir 17 to the water distribution reservoir 18.

[0022] The water distribution reservoir 18 temporarily stores the treated water supplied from the water purification reservoir 17 and supplies it to households, etc. (not shown).

[0023] Storage tank T is, for example, a tank for storing a coagulant to be injected into the water to be treated.

[0024] Pump P3 is, for example, a pump installed in the piping connecting the storage tank T and the mixing tank 13. Specifically, pump P3 supplies a coagulant to the mixing tank 13 at an injection rate predetermined by the administrator.

[0025] Furthermore, the water treatment equipment 100 may also include, for example, other pumps (not shown) besides pumps P1, P2, and P3. Additionally, the water treatment equipment 100 may supply chemicals other than coagulants (for example, caustic soda) to the water to be treated.

[0026] The information processing device 1 is, for example, an electronic device having an electronic circuit. Specifically, the information processing device 1 is, for example, one or more physical machines or one or more virtual machines having a CPU (Central Processing Unit) and memory.

[0027] The information processing device 1 then performs a process (hereinafter also called the model generation process) to generate a learning model (hereinafter also called the learning model) that predicts time-series data of water volume information (hereinafter also called the water volume information) indicating the amount of treated water that needs to be supplied to homes, etc., by using, for example, meteorological information (hereinafter also simply called meteorological information) indicating weather conditions.

[0028] Furthermore, the information processing device 1 performs, for example, a process to predict time-series data of water volume information using a learning model (hereinafter also referred to as water demand forecasting).

[0029] Specifically, in the model generation process, the information processing device 1 generates a learning model (hereinafter referred to as the first learning model) by performing machine learning on multiple training data sets (hereinafter also referred to as the first training data) corresponding to multiple time periods (hereinafter also referred to as the first time periods), each of which includes meteorological information (hereinafter also referred to as the first meteorological information) indicating the weather conditions in a predetermined region (hereinafter also referred to as the first region) for each time period, and time-series data of water volume information (hereinafter also referred to as the first water volume information) indicating the amount of treated water needed in the first region for one or more time periods corresponding to each time period (one or more time periods after each time period). The information processing device 1 then stores the generated first learning model in a memory unit. Note that each of the multiple first time periods is, for example, a past time period.

[0030] Subsequently, in the water demand forecasting process, the information processing device 1 inputs, for example, meteorological information (hereinafter also referred to as second meteorological information) indicating weather conditions in other time zones (hereinafter also referred to as second time zones) and other regions (hereinafter also referred to as second regions) to the first learning model stored in the memory unit. Then, the information processing device 1 acquires, for example, water volume information (hereinafter also referred to as second water volume information) indicating the amount of treated water required in the second region for one or more time zones corresponding to the second time zone (one or more time zones after the second time zone), which is output (predicted) in conjunction with the input of the second meteorological information to the first learning model. Furthermore, the information processing device 1 outputs, for example, the acquired second water volume information. Note that the second region may be, for example, the same region as the first region, or a different region from the first region. Also, the second time zone may be, for example, a past or present time zone, or a future time zone.

[0031] In other words, changes in water volume information at a time period later than a specific time period in a particular region can be determined to be correlated with, for example, weather information at a specific time period in that region. Weather information includes, for example, information indicating the weather, temperature, and precipitation at a specific time period in a particular region.

[0032] Therefore, the information processing device 1 in this embodiment can accurately predict time-series data showing the changes in water volume information in time periods later than other time periods (second time periods) by using a learning model (first learning model) generated by machine learning of multiple training data (first training data) which include weather information (first weather information) and water volume information (first water volume information) associated with each of multiple time periods (first time periods). Specifically, in the water demand forecasting process, the information processing device 1 in this embodiment can accurately predict time-series data of water volume information in time periods later than other time periods by inputting weather information (second weather information) from other time periods into the learning model.

[0033] As a result, the information processing device 1 in this embodiment can accurately predict, for example, the amount of purified water that needs to be supplied to households, or in other words, the amount of purified water that needs to be generated in the water treatment facility 100.

[0034] Returning to Figure 1, the control device 2 is, for example, an electronic device having an electronic circuit. Specifically, the control device 2 is, for example, one or more physical machines or one or more virtual machines having a CPU and memory.

[0035] The control device 2 then controls each water treatment facility 100 based, for example, on the results of the water demand forecasting process performed by the information processing device 1.

[0036] Specifically, the control device 2 controls the amount of treated water (amount per unit time) supplied from the water purification reservoir 17 to the water distribution reservoir 18 by controlling the opening degree of a valve (not shown) provided on the pump P2, for example, based on the prediction results from the information processing device 1 (prediction results for time-series data of water volume information).

[0037] Furthermore, if, for example, multiple pumps including pump P2 (or simply multiple pumps) are installed in the piping connecting the water purification reservoir 17 and the water distribution reservoir 18, the control device 2 may calculate the amount of treated water (amount per unit time) that needs to be discharged from the multiple pumps based on the prediction results from the information processing device 1, and then control the amount of treated water supplied to the water distribution reservoir 18 by operating the necessary number of pumps to discharge the amount of treated water corresponding to the calculated amount.

[0038] [Information processing device 1 in the first embodiment] Next, the configuration of the information processing device 1 in the first embodiment will be described. Figure 3 is a diagram illustrating the hardware configuration of the information processing device 1.

[0039] As shown in Figure 3, the information processing device 1 is a computer device having, for example, a CPU 101 which is a processor, memory 102, a communication device 103, and a storage medium 104. Each part is connected to the others, for example, via a bus 105.

[0040] The storage medium 104 has, for example, a program storage area (not shown) for storing a program 110 for performing model learning processing and water demand forecasting processing (hereinafter collectively referred to simply as water demand forecasting processing, etc.). The storage medium 104 also has, for example, an information storage area 130 for storing information used when performing water demand forecasting processing, etc. The storage medium 104 may be, for example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive).

[0041] The CPU 101 performs tasks such as water demand forecasting by executing a program 110 loaded into memory 102 from storage medium 104.

[0042] The communication device 103 accesses, for example, the control device 2, an operating terminal (not shown) where the operator inputs necessary information, and other information processing devices (not shown) capable of providing various types of information used for water demand forecasting, etc., via a network (not shown) such as Ethernet.

[0043] The information processing device 1 may, for example, have an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). Furthermore, water demand forecasting and the like may be executed, for example, on the FPGA or ASIC.

[0044] Furthermore, the following description will assume that the water treatment system 1000 has one information processing device 1, but it is not limited to this. Specifically, the water treatment system 1000 may have, for example, multiple information processing devices 1. The model learning process and the water demand forecasting process may each be performed on different information processing devices 1, for example. Moreover, the model learning process may be performed in a distributed manner across multiple information processing devices 1, for example. Similarly, the water demand forecasting process may also be performed in a distributed manner across multiple information processing devices 1, for example.

[0045] [Functions of Information Processing Device 1] Next, the functions of the information processing device 1 will be described. Figure 4 is a block diagram of the functions of the information processing device 1 in the first embodiment. Figures 5 and 6 illustrate specific examples of the model learning process in the first embodiment. Figures 7 and 8 illustrate specific examples of the water demand forecasting process in the first embodiment.

[0046] As shown in Figure 4, the information processing device 1 implements various functions, including an information acquisition unit 111, an information management unit 112, a data generation unit 113, a model generation unit 114, an information estimation unit 115, and an information output unit 116, through the organic cooperation of hardware such as the CPU 101 and memory 102 with a program.

[0047] Furthermore, as shown in Figure 4, the information processing device 1 stores, for example, weather information DT1 (hereinafter also referred to as first weather information DT1), water volume information DT2 (hereinafter also referred to as water volume information DT2), training data DT11 (hereinafter also referred to as first training data DT11), training data DT12 (hereinafter also referred to as second training data DT12), learning model MD1 (hereinafter also referred to as first learning model MD1), and learning model MD2 (hereinafter also referred to as second learning model MD2) in the information storage area 130.

[0048] First, we will explain the functions of the model learning process.

[0049] The information acquisition unit 111 acquires weather information DT1, which is past weather information (actual information) generated (aggregated) by another first information processing system (not shown), at first timings such as every hour (hereinafter also simply referred to as the first timing). Weather information DT1 is, for example, weather information for the first region during the time period from the previous first timing to the current first timing. Weather information DT1 also includes, for example, information indicating the time period from the previous first timing to the current first timing (hereinafter also referred to as time information).

[0050] Specifically, the information acquisition unit 111 receives, for example, weather information DT1 transmitted by another first information processing system. The information acquisition unit 111 also acquires, for example, weather information DT1 received from another first information processing system by another information processing device (not shown) of the water treatment system 1000. The information acquisition unit 111 also acquires, for example, weather information DT1 by accessing another first information processing system. The information acquisition unit 111 also acquires, for example, weather information DT1 manually entered by an operator. The information management unit 112 then stores, for example, the weather information DT1 acquired by the information acquisition unit 111 in the information storage area 130.

[0051] Furthermore, the information acquisition unit 111 may repeatedly acquire weather information DT1 for the same time period if the information is such that the weather information DT1 generated by another first information processing system may be updated.The information management unit 112 may then update the repeatedly acquired weather information DT1 from the information stored in the information storage area 130 if the information acquisition unit 111 has repeatedly acquired weather information DT1 for the same time period.

[0052] Furthermore, the information acquisition unit 111 acquires, for example, water volume information DT2, which is past water volume information (actual information) generated (aggregated) by another second information processing system (not shown), at each first timing. Water volume information DT2 is, for example, water volume information for the first region during the time period from the previous first timing to the current first timing. Note that water volume information DT2 also includes, for example, time information indicating the time period from the previous first timing to the current first timing.

[0053] Specifically, the information acquisition unit 111 receives, for example, water volume information DT2 transmitted by another second information processing system. The information acquisition unit 111 also acquires, for example, water volume information DT2 received from another second information processing system by another information processing device of the water treatment system 1000. The information acquisition unit 111 also acquires, for example, water volume information DT2 by accessing another second information processing system. The information acquisition unit 111 also acquires, for example, water volume information DT2 manually entered by an operator. The information management unit 112 then stores, for example, the water volume information DT2 acquired by the information acquisition unit 111 in the information storage area 130.

[0054] Furthermore, if the information acquisition unit 111 is information that may be updated by another second information processing system, for example, it may repeatedly acquire the water volume information DT2 for the same time period.Then, if the information acquisition unit 111 repeatedly acquires the water volume information DT2 for the same time period, the information management unit 112 may update the information stored in the information storage area 130 with the repeatedly acquired water volume information DT2.

[0055] As shown in Figure 5, the data generation unit 113 generates multiple training data sets DT11, each containing, for example, time-series data of weather information DT1 and water volume information DT2 acquired by the information acquisition unit 111. The information management unit 112 then stores the multiple training data sets DT11 generated by the data generation unit 113 in the information storage area 130.

[0056] Specifically, the data generation unit 113 generates multiple training data sets DT11 for each of several time periods, each containing weather information DT1 corresponding to that time period and time-series data of water volume information DT2 for one or more time periods after that time period.

[0057] Furthermore, as shown in Figure 6, the data generation unit 113 generates multiple training data sets DT12, each containing time-series data of water volume information DT2 acquired by the information acquisition unit 111. That is, each of the multiple training data sets DT12 is different from each of the multiple training data sets DT11, for example, and does not contain weather information DT1. The information management unit 112 then stores the multiple training data sets DT12 generated by the data generation unit 113 in the information storage area 130.

[0058] Specifically, the data generation unit 113 generates, for example, multiple training data sets DT12 for each of several time periods, each containing time-series data of water volume information DT2 for one or more time periods after each respective time period.

[0059] As shown in Figure 5, the model generation unit 114 generates a learning model MD1 by, for example, performing machine learning on multiple training data DT11 generated by the data generation unit 113. The information management unit 112 then stores the learning model MD1 generated by the model generation unit 114 in the information storage area 130.

[0060] Furthermore, the data generation unit 113 may, for example, generate new training data DT11 that includes the new weather information DT1 and new water volume information DT2 each time the information acquisition unit 111 acquires new weather information DT1 and new water volume information DT2. The model generation unit 114 may, for example, regenerate (update) the learning model MD1 by retraining multiple training data DT11 that include the new training data DT11. Specifically, the model generation unit 114 may, for example, regenerate the learning model MD1 by retraining multiple training data DT11 that include the new training data DT11 at regular intervals. This makes it possible for the information processing device 1 to further improve the judgment accuracy of the learning model MD1 by having the learning model MD1 learn new training data DT11 that includes the most recent weather information DT1 and the most recent water volume information DT2.

[0061] Furthermore, as shown in Figure 6, the model generation unit 114 generates a learning model MD2 by, for example, performing machine learning on multiple training data DT12 generated by the data generation unit 113. The information management unit 112 then stores the learning model MD2 generated by the model generation unit 114 in the information storage area 130.

[0062] Furthermore, the data generation unit 113 may, for example, generate new training data DT12 including the new water volume information DT2 each time the information acquisition unit 111 acquires new water volume information DT2. The model generation unit 114 may, for example, regenerate (update) the learning model MD2 by retraining multiple training data DT12 including the new training data DT12. Specifically, the model generation unit 114 may, for example, regenerate the learning model MD2 by retraining multiple training data DT12 including the new training data DT12 at regular intervals. This makes it possible for the information processing device 1 to further improve the judgment accuracy of the learning model MD2 by having the learning model MD2 learn the new training data DT12 including the most recent water volume information DT2.

[0063] Next, we will explain the functions of the water demand forecasting process.

[0064] The information acquisition unit 111 acquires weather information DT1 (hereinafter also referred to as weather information DT1a or second weather information DT1a), which is future weather information (prediction information) generated (aggregated) by another third information processing system (not shown), at a second timing (hereinafter also simply referred to as the second timing), for example, every hour. Weather information DT1a is, for example, the weather information for the second region for each time period from the current second timing to a predetermined time later.

[0065] Specifically, the information acquisition unit 111 receives, for example, weather information DT1a transmitted by another third information processing system. The information acquisition unit 111 also acquires, for example, weather information DT1a received from another third information processing system by another information processing device of the water treatment system 1000. The information acquisition unit 111 also acquires, for example, weather information DT1a by accessing another third information processing system. The information acquisition unit 111 also acquires, for example, weather information DT1a manually entered by an operator. The information management unit 112 then stores, for example, the weather information DT1a acquired by the information acquisition unit 111 in the information storage area 130.

[0066] Furthermore, if the information acquisition unit 111 is unable to acquire weather information DT1a from another third information processing system due to, for example, a communication failure between the other third information processing system and the information processing device 1, it will not acquire weather information DT1a from the other third information processing system.

[0067] The information estimation unit 115 determines, for example, at each second timing, whether the elapsed time since the last time the information acquisition unit 111 acquired the weather information DT1a exceeds a predetermined threshold (hereinafter also simply referred to as the threshold). That is, for example, at each second timing, the information estimation unit 115 determines whether the elapsed time since the last time the information acquisition unit 111 acquired the weather information DT1a exceeds a threshold due to a communication failure or the like between another third information processing system and the information processing device 1.

[0068] As a result, for example, if the information acquisition unit 111 determines that the elapsed time since the last acquisition of weather information DT1a does not exceed a threshold, the information estimation unit 115 inputs the weather information DT1a stored in the information storage area 130 (for example, the latest weather information DT1a among the weather information DT1a stored in the information storage area 130) to the learning model MD1 generated by the model generation unit 114, as shown in Figure 7. Subsequently, the information estimation unit 115 acquires, for example, the time-series data of the water volume information DT2 (hereinafter also referred to as water volume information DT2a or second water volume information DT2a) that is output (predicted) in conjunction with the input of weather information DT1a to the learning model MD1.

[0069] In other words, if the elapsed time since the last time the information acquisition unit 111 acquired the weather information DT1a does not exceed a threshold, it means, for example, that the weather information DT1a stored in the information storage area 130 can be judged to have predicted the weather conditions at the current second timing with sufficient accuracy, and that even when used as input data for the learning model MD1, it can be judged not to cause a decrease in the estimation accuracy of the learning model MD1. Therefore, in this case, the information estimation unit 115 uses, for example, the learning model MD1 which was generated on the premise that the weather information DT1a is input, to estimate the time series data of water volume information DT2a for one or more time periods after the time period corresponding to the weather information DT1a.

[0070] On the other hand, if the information acquisition unit 111 determines, for example, that the elapsed time since the last acquisition of weather information DT1a exceeds a threshold, the information estimation unit 115 inputs time information DT0, which indicates the current second timing, to the learning model MD2 generated by the model generation unit 114, as shown in Figure 8. Subsequently, the information estimation unit 115 acquires time-series data of water volume information DT2 (hereinafter also referred to as water volume information DT2b or other second water volume information DT2b) that is output (predicted) in conjunction with the input of time information DT0 to the learning model MD2.

[0071] In other words, if the elapsed time since the last time the information acquisition unit 111 acquired the weather information DT1a exceeds a threshold, it means, for example, that the weather information DT1a stored in the information storage area 130 does not predict the weather conditions at the current second timing with sufficient accuracy, and that using it as input data for the learning model MD1 may cause a decrease in the estimation accuracy of the learning model MD1. Therefore, in this case, the information estimation unit 115 uses, for example, the learning model MD2, which was generated assuming that the weather information DT1a is not input, instead of the learning model MD1, which was generated assuming that the weather information DT1a is input, to estimate the time series data of water volume information DT2b for one or more time periods after the time period corresponding to the weather information DT1a.

[0072] The information output unit 116 outputs, for example, time-series data of water volume information DT2a or water volume information DT2b acquired by the information estimation unit 115. Specifically, the information output unit 116 transmits, for example, the time-series data of water volume information DT2a or water volume information DT2b acquired by the information estimation unit 115 to the control device 2. Subsequently, the control device 2 controls each water treatment facility 100, for example, based on the time-series data of water volume information DT2a or water volume information DT2b transmitted from the information processing device 1.

[0073] [Model learning process in the first embodiment] Next, the model learning process in the first embodiment will be described. Figure 9 is a flowchart illustrating the model learning process in the first embodiment. Figures 10 to 13 are diagrams illustrating the model learning process in the first embodiment.

[0074] The information acquisition unit 111 acquires weather information DT1, which is past performance information generated (aggregated) by another first information processing system, at each first timing (step S1 in Figure 9). The information management unit 112 then stores the weather information DT1 acquired in step S1 in the information storage area 130. A specific example of weather information DT1 will be described below.

[0075] [Specific example of weather information DT1] Figure 10 is a diagram illustrating a specific example of weather information DT1.

[0076] As shown in Figure 10, the weather information DT1 has items such as "Date" which sets the date to which each time period is included, "Time Period" which sets each time period, "Temperature" which sets the temperature for each time period, "Weather" which sets the weather for each time period, and "Precipitation" which sets the amount of precipitation for each time period.

[0077] Specifically, in the weather information DT1 shown in Figure 10, the information in the first row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "0:00-1:00", "Temperature" set to "18 (degrees)", "Weather" set to "Sunny", and "Precipitation" set to "0 (mm)".

[0078] In other words, the information in the first line of the weather information DT1 shown in Figure 10 indicates, for example, that the temperature was 18 degrees Celsius, the weather was sunny, and the precipitation was 0 mm during the period from 0:00 to 1:00 on July 1st.

[0079] Furthermore, in the weather information DT1 shown in Figure 10, the information in the second row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "1:00-2:00", "Temperature" set to "16 (degrees)", "Weather" set to "Cloudy", and "Precipitation" set to "0 (mm)".

[0080] In other words, the information in the second line of the weather information DT1 shown in Figure 10 indicates, for example, that the temperature was 16 degrees Celsius, the weather was cloudy, and the precipitation was 0 mm during the period from 1:00 to 2:00 on July 1st. Explanations of the other information included in Figure 10 are omitted.

[0081] Furthermore, the information acquisition unit 111 may process, for example, at least a portion of the weather information DT1 acquired from another first information processing system. Specifically, if the information acquisition unit 111 does not contain information indicating the weather for each time period in the weather information DT1 acquired from another first information processing system, it may estimate the information indicating the weather for each time period from information indicating the temperature for each time period and information indicating the amount of precipitation for each time period, and include the estimated information (information indicating the weather for each time period) in the weather information DT1.

[0082] Returning to Figure 9, the information acquisition unit 111 acquires, for example, water volume information DT2, which is past performance information generated (aggregated) by another second information processing system, at each first timing (step S2 in Figure 9). The information management unit 112 then stores, for example, the water volume information DT2 acquired in step S2 in the information storage area 130. A specific example of water volume information DT2 will be explained below.

[0083] [Specific example of water volume information DT2] Figure 11 illustrates a specific example of water volume information DT2.

[0084] As shown in Figure 11, the water volume information DT2 has items such as "Date," which sets the date to which each time period is included; "Time Period," which sets each time period; and "Water Demand," which sets the amount of treated water supplied to households, etc., during each time period.

[0085] Specifically, in the water volume information DT2 shown in Figure 11, the information in the first row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "0:00-1:00", and "Water Demand" set to "80(m 3 The setting is " / h)".

[0086] In other words, the first row of the water volume information DT2 shown in Figure 11 indicates, for example, that the amount of treated water supplied to households, etc., during the period from 0:00 to 1:00 on July 1st was 80 (m³). 3 This indicates that it was / h).

[0087] Furthermore, in the water volume information DT2 shown in Figure 11, the information in the second row is, for example, set as "Date" as "7 / 1", set as "Time Zone" as "1:00-2:00", and set as "Water Demand" as "50(m 3 The setting is " / h)".

[0088] In other words, the information in the second row of the water volume information DT2 shown in Figure 11 indicates, for example, that the amount of treated water supplied to households, etc., during the time period from 1:00 to 2:00 on July 1st was 50 (m³). 3 This indicates that it was / h). Explanations of other information included in Figure 11 are omitted.

[0089] Returning to Figure 9, the data generation unit 113 generates multiple training data sets DT11, each containing, for example, time-series data of weather information DT1 and water volume information DT2 stored in the information storage area 130 (step S3 in Figure 9). The information management unit 112 then stores, for example, the multiple training data sets DT11 generated in step S3 in the information storage area 130. A specific example of the training data sets DT11 will be described below.

[0090] [Specific example of training data DT11] Figure 12 illustrates a specific example of the training data DT11.

[0091] As shown in Figure 12, the training data DT11 has items such as "Date" which sets the date to which each time period is included, "Time Period" which sets each time period, "Temperature" which sets the temperature in each time period, "Weather" which sets the weather in each time period, "Precipitation" which sets the amount of precipitation in each time period, "Water Demand (1 hour later)" which sets the amount of treated water that needed to be supplied to households, etc. in the time period one hour later of each time period, "Water Demand (2 hours later)" which sets the amount of treated water that needed to be supplied to households, etc. in the time period two hours later of each time period, and "Water Demand (3 hours later)" which sets the amount of treated water that needed to be supplied to households, etc. in the time period three hours later of each time period.

[0092] Specifically, in the training data DT11 shown in Figure 12, the information in the first row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "0:00-1:00", "Temperature" set to "18 (degrees)", "Weather" set to "Sunny", "Precipitation" set to "0 (mm)", and "Water Demand (1 hour later)" set to "50 (m³)". 3 The value " / h)" is set to "Water demand (after 2 hours)" and "50(m 3 The value " / h)" is set to "Water demand (after 3 hours)" and "40(m 3 The setting is " / h)".

[0093] In other words, the information in the first row of the training data DT11 shown in Figure 12 includes, for example, the information set for "temperature," "weather," and "precipitation" in the first row of the weather information DT1 shown in Figure 10, the information set for "water demand" in the second row of the water volume information DT2 shown in Figure 11, the information set for "water demand" in the third row of the water volume information DT2 shown in Figure 11, and the information set for "water demand" in the fourth row of the water volume information DT2 shown in Figure 11.

[0094] Also, in the teacher data DT11 shown in FIG. 12, in the information on the second line, for example, "7 / 1" is set as the "date", "1:00 - 2:00" is set as the "time zone", "16 (degrees)" is set as the "temperature", "cloudy" is set as the "weather", "0 (mm)" is set as the "precipitation amount", "water demand (1 hour later)" is set as "50 (m 3 / h)", "water demand (2 hours later)" is set as "40 (m 3 / h)", and "water demand (3 hours later)" is set as "30 (m 3 / h)".

[0095] That is, the information on the second line in the teacher data DT11 shown in FIG. 12 includes, for example, the information set for "temperature", "weather", and "precipitation amount" in the information on the second line in the meteorological information DT1 shown in FIG. 10, the information set for "water demand" in the information on the third line in the water volume information DT2 shown in FIG. 11, the information set for "water demand" in the information on the fourth line in the water volume information DT2 shown in FIG. 11, and the information set for "water demand" in the information on the fifth line in the water volume information DT2 shown in FIG. 11. Explanation of other data included in FIG. 12 is omitted.

[0096] Note that the teacher data DT11 may have, for example, "water demand" in which the amount of treated water that needs to be supplied to a home or the like in each time zone is set, instead of the amount of treated water that needs to be supplied to a home or the like in the time zone several hours after each time zone such as "water demand (1 hour later)", "water demand (2 hours later)", and "water demand (3 hours later)".

[0097] Returning to FIG. 9, the data generation unit 113 generates, for example, a plurality of teacher data DT12 each including the time - series data of the water volume information DT2 stored in the information storage area 130 (step S4 in FIG. 9). Then, the information management unit 112 stores, for example, the plurality of teacher data DT12 generated in step S4 in the information storage area 130. Hereinafter, a specific example of the teacher data DT12 will be described.

[0098] [Specific Example of Teacher Data DT12] Figure 13 illustrates a specific example of the training data DT12.

[0099] As shown in Figure 13, the training data DT12 has the following items: "Date" which sets the date to which each time period is included; "Time Period" which sets each time period; "Water Demand (1 hour later)" which sets the amount of treated water that needed to be supplied to households, etc., one hour after each time period; "Water Demand (2 hours later)" which sets the amount of treated water that needed to be supplied to households, etc., two hours after each time period; and "Water Demand (3 hours later)" which sets the amount of treated water that needed to be supplied to households, etc., three hours after each time period.

[0100] Specifically, in the training data DT12 shown in Figure 13, the information in the first row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "0:00-1:00", and "Water Demand (1 hour later)" set to "50 (m 3 The value " / h)" is set to "Water demand (after 2 hours)" and "50(m 3 The value " / h)" is set to "Water demand (after 3 hours)" and "40(m 3 The setting is " / h)".

[0101] In other words, the information in the first row of the training data DT12 shown in Figure 13 includes, for example, the information set as "water demand" in the second row of the water volume information DT2 shown in Figure 11, the information set as "water demand" in the third row of the water volume information DT2 shown in Figure 11, and the information set as "water demand" in the fourth row of the water volume information DT2 shown in Figure 11.

[0102] Furthermore, in the training data DT12 shown in Figure 13, the information in the second row includes, for example, "Date" set to "7 / 1", "Time Zone" set to "1:00-2:00", and "Water Demand (1 hour later)" set to "50 (m 3 The value " / h)" is set to "Water demand (after 2 hours)" and "40(m 3 The value " / h)" is set to "Water demand (after 3 hours)" and "30(m3 The setting is " / h)".

[0103] In other words, the information in the second row of the training data DT12 shown in Figure 13 includes, for example, the information set as "water demand" in the third row of the water volume information DT2 shown in Figure 11, the information set as "water demand" in the fourth row of the water volume information DT2 shown in Figure 11, and the information set as "water demand" in the fifth row of the water volume information DT2 shown in Figure 11. The explanation of the other data included in Figure 13 is omitted.

[0104] Furthermore, the training data DT12 may include an item called "Water Demand," which sets the amount of treated water that needed to be supplied to households, etc., at each time period, instead of the amount of treated water that needed to be supplied to households, etc., at each time period several hours later, such as "Water Demand (1 hour later)," "Water Demand (2 hours later)," and "Water Demand (3 hours later)."

[0105] Returning to Figure 9, the model generation unit 114 generates a learning model MD1 by, for example, performing machine learning on multiple training data DT11 stored in the information storage area 130 (step S5 in Figure 9). Then, the information management unit 112 stores the learning model MD1 generated in step S5 in the information storage area 130.

[0106] Furthermore, the model generation unit 114 generates a learning model MD2 by, for example, performing machine learning on multiple training data DT12 stored in the information storage area 130 (step S6 in Figure 9). Then, the information management unit 112 stores the learning model MD2 generated in step S6 in the information storage area 130.

[0107] In the example above, we described the case where the creation of learning model MD1 and learning model MD2 are performed sequentially, but this is not the only case. Specifically, the information processing device 1 may, for example, perform the creation of learning model MD1 and learning model MD2 in parallel. Also, the information processing device 1 may, for example, perform the creation of learning model MD2 before the creation of learning model MD1.

[0108] [Water demand forecasting process in the first embodiment] Next, the water demand forecasting process in the first embodiment will be described. Figure 14 is a flowchart illustrating the water demand forecasting process in the first embodiment. Figures 15 to 18 are diagrams illustrating the water demand forecasting process in the first embodiment.

[0109] The information acquisition unit 111 acquires weather information DT1a, which is future forecast information generated (aggregated) by another third information processing system, at each second timing (step S11 in Figure 14). The information management unit 112 then stores the weather information DT1a acquired in step S11 in the information storage area 130.

[0110] Furthermore, if the information acquisition unit 111 is unable to acquire weather information DT1a from another third information processing system due to, for example, a communication failure between the other third information processing system and the information processing device 1, step S11 may be omitted. Specific examples of weather information DT1a will be described below.

[0111] [Specific example of weather information DT1a] Figure 15 illustrates a specific example of weather information DT1a.

[0112] As shown in Figure 15, the weather information DT1a has the following items: "Date," which sets the date that includes the time period (future time period) corresponding to the weather information DT1a; "Time Period," which sets the time period corresponding to the weather information DT1a; "Temperature," which sets the temperature (predicted information) for the time period corresponding to the weather information DT1a; "Weather," which sets the weather (predicted information) for the time period corresponding to the weather information DT1a; and "Precipitation," which sets the precipitation (predicted information) for the time period corresponding to the weather information DT1a.

[0113] Specifically, in the weather information DT1a shown in Figure 15, for example, "8 / 1" is set as the "Date," "0:00-1:00" as the "Time Zone," "21 (degrees)" as the "Temperature," "Rain" as the "Weather," and "3 (mm)" as the "Precipitation."

[0114] Returning to Figure 14, the information estimation unit 115 determines, for example, at each second timing, whether the elapsed time since the acquisition of the latest weather information DT1a in step S11 exceeds a threshold (step S12 in Figure 14).

[0115] As a result, for example, if it is determined in step S11 that the elapsed time since the acquisition of the latest weather information DT1a does not exceed a threshold (YES in step S12 in Figure 14), the information estimation unit 115 inputs the weather information DT1a acquired in step S11 to the learning model MD1 stored in the information storage area 130 (step S13 in Figure 14).

[0116] Subsequently, the information estimation unit 115 acquires, for example, time-series data of water volume information DT2a that is output (predicted) in response to the input of weather information DT1a to the learning model MD1. Then, the information output unit 116 outputs, for example, the acquired time-series data of water volume information DT2a to the control device 2 (step S14 in Figure 14). A specific example of the time-series data of water volume information DT2a will be described below.

[0117] [Specific example of time-series data from water volume information DT2a] Figure 16 illustrates a specific example of time-series data for water volume information DT2a.

[0118] As shown in Figure 16, the time-series data of water volume information DT2a includes items such as "Water Demand (1 hour later)," which sets the amount of treated water that needs to be supplied to households, etc., one hour after the time period corresponding to weather information DT1a; "Water Demand (2 hours later)," which sets the amount of treated water that needs to be supplied to households, etc., two hours after the time period corresponding to weather information DT1a; and "Water Demand (3 hours later)," which sets the amount of treated water that needs to be supplied to households, etc., three hours after the time period corresponding to weather information DT1a.

[0119] Specifically, the time-series data of water volume information DT2a shown in Figure 16 includes, for example, "Water demand (1 hour later)" as "40 (m³ 3 The value " / h)" is set to "Water demand (after 2 hours)" and "30(m 3 The value " / h)" is set to "Water demand (after 3 hours)" and "20(m 3 The setting is " / h)". Further explanation of the other information included in Figure 16 is omitted.

[0120] Returning to Figure 14, for example, if step S11 determines that the elapsed time since the acquisition of the latest weather information DT1a exceeds a threshold (NO in step S12 of Figure 14), the information estimation unit 115 inputs time information DT0 corresponding to the current second timing to the learning model MD2 stored in the information storage area 130 (step S15 of Figure 14). A specific example of time information DT0 will be explained below.

[0121] [Specific example of time information DT0] Figure 17 illustrates a specific example of time information DT0.

[0122] As shown in Figure 17, the time information DT0 has items such as "Date," which is set to the date that includes the time period corresponding to the time information DT0, and "Time Period," which is set to the time period corresponding to the time information DT0.

[0123] Specifically, the time information DT0 shown in Figure 17 has, for example, "8 / 1" set as the "date" and "0:00-1:00" set as the "time zone".

[0124] Returning to Figure 14, the information estimation unit 115 acquires, for example, time-series data of the water volume information DT2b that is output (predicted) in response to the input of time information DT0 to the learning model MD2. Then, the information output unit 116 outputs, for example, the acquired time-series data of the water volume information DT2b (step S16 in Figure 14). A specific example of the time-series data of the water volume information DT2b will be explained below.

[0125] [Specific example of time-series data from water volume information DT2b] Figure 18 illustrates a specific example of time-series data for water volume information DT2b.

[0126] The time-series data of water volume information DT2b has the same items as, for example, the water volume information DT2a described in Figure 16, as shown in Figure 18.

[0127] Specifically, the time-series data of water volume information DT2b shown in Figure 18 includes, for example, "Water demand (1 hour later)" as "40 (m³ 3 The value " / h)" is set to "Water demand (after 2 hours)" and "25(m 3 The value " / h)" is set to "Water demand (after 3 hours)" and "20(m 3 The setting is " / h)". Further explanation of the other information included in Figure 18 is omitted.

[0128] As described above, the information processing device 1 in this embodiment generates a first learning model MD1 by performing machine learning on a plurality of first training data DT11, each of which corresponds to a plurality of first time zones, and which each includes a first weather information DT1 indicating the weather conditions in a first region for each time zone, and time-series data of first water volume information DT2 indicating the amount of treated water needed in the first region for one or more time zones corresponding to each time zone. The information processing device 1 in this embodiment then stores the generated first learning model MD1 in the information storage area 130.

[0129] Specifically, the first weather information DT1 includes at least one of the following: information indicating the weather in the first region for each time period, information indicating the temperature in the first region for each time period, and information indicating the amount of precipitation in the first region for each time period.

[0130] In this embodiment, the information processing device 1 generates a second learning model MD2 by performing machine learning on a plurality of second training data DT12, each corresponding to a plurality of first time periods, which does not include the first weather information DT1 corresponding to each time period, but each includes time-series data of the first water volume information DT2 corresponding to each time period. The information processing device 1 then stores the generated second learning model MD2 in the information storage area 130.

[0131] Furthermore, the information processing device 1 in this embodiment, for example, inputs second weather information DT1a indicating the weather conditions in the second region during the second time period to the first learning model MD1 stored in the information storage area 130, and acquires time-series data of second water volume information DT2a indicating the amount of treated water required in the second region for one or more time periods corresponding to the second time period, which is output in conjunction with the input of the second weather information DT1a to the first learning model MD1.Then, the information processing device 1 in this embodiment outputs, for example, the acquired time-series data of second water volume information DT2a.

[0132] Specifically, in this embodiment, the information processing device 1 inputs the second weather information DT1a to the first learning model MD1 stored in the information storage area 130 if the second weather information DT1a satisfies predetermined conditions. The predetermined conditions are, for example, when the elapsed time since the information processing device 1 last acquired the second weather information DT1a does not exceed a threshold.

[0133] On the other hand, in this embodiment, if the second weather information DT1a does not meet predetermined conditions, the information processing device 1 inputs time information DT0 indicating the second time zone to the second learning model MD2 stored in the information storage area 130. Then, in this embodiment, the information processing device 1 acquires time-series data of other second water volume information DT2b, which indicates the amount of treated water needed in the second region for one or more time zones corresponding to the second time zone, and which is output in conjunction with the input of time information DT0 indicating the second time zone to the second learning model MD2.

[0134] Subsequently, the information processing device 1 in this embodiment outputs, for example, time-series data of the acquired second water volume information DT2a or time-series data of the other second water volume information DT2b.

[0135] As a result, the information processing device 1 in this embodiment can accurately predict, for example, the amount of purified water that needs to be supplied to households, or in other words, the amount of purified water that needs to be generated in the water treatment facility 100.

[0136] Furthermore, the information processing device 1 in this embodiment can reduce the amount of manual work required, such as inputting various types of information (e.g., weather information) into a mathematical formula, compared to a method that predicts water volume information using a predetermined mathematical formula (hereinafter simply referred to as a formula). Therefore, the information processing device 1 in this embodiment can reduce the burden on workers associated with predicting water volume information.

[0137] Furthermore, the information processing device 1 in this embodiment can, for example, automatically acquire weather information DT1a and input it into the learning model MD1. As a result, the information processing device 1 in this embodiment can, for example, predict water volume information DT2a at desired intervals, and make predictions that follow changes in temperature, weather, etc. Consequently, the information processing device 1 in this embodiment can, for example, output water volume information DT2a with higher prediction accuracy to the control device 2, making it possible to control the water treatment equipment 100 with greater precision.

[0138] Furthermore, weather information DT1, weather information DT1a, and training data DT11 may include, for example, at least one of the following: the average temperature over the most recent few hours for each time period, the maximum temperature over the most recent few hours for each time period, the average precipitation over the most recent few hours for each time period, and the maximum precipitation over the most recent few hours for each time period.

[0139] Furthermore, weather information DT1, weather information DT1a, training data DT11, and training data DT12 may include, for example, at least one of the following: the day of the week in which each time period is included; a flag indicating whether the day in which each time period is included is a public holiday; a flag indicating whether the day before the day in which each time period is included is a public holiday; a flag indicating whether the day after the day in which each time period is included is a public holiday; a flag indicating whether the day in which each time period is included is the day of an event, etc.; a flag indicating whether the day before the day in which each time period is included is the day of an event, etc.; and a flag indicating whether the day after the day in which each time period is included is the day of an event, etc.

[0140] Furthermore, the training data DT11 may include, for example, time-series data of water volume information DT2 for one or more time periods prior to the time period corresponding to the weather information DT1. In this case, the information estimation unit 115 may, for example, input not only the weather information DT1a but also the time-series data of water volume information DT2 for one or more time periods prior to the time period corresponding to the weather information DT1a to the learning model MD1, thereby obtaining time-series data of water volume information DT2a for one or more time periods after the time period corresponding to the weather information DT1a.

[0141] Furthermore, the training data DT12 may include, for example, time-series data of water volume information DT2 for one or more time periods prior to the time period corresponding to the time information DT0. In this case, the information estimation unit 115 may, for example, input not only the time information DT0 but also the time-series data of water volume information DT2 for one or more time periods prior to the time period corresponding to the time information DT0 to the learning model MD2, thereby obtaining time-series data of water volume information DT2b for one or more time periods after the time period corresponding to the time information DT0. [Explanation of Symbols]

[0142] 1: Information processing device 2: Control device 11: Sand basin 12: Landing well 13: Mixing pond 14: Flocculation pond 15: Sedimentation tank 16: Filtration tank 17: Water purification reservoir 18: Water distribution reservoir 100: Water treatment equipment 101: CPU 102: Memory 103: Communication device 104: Storage medium 105: Bus 110: Program 111: Information Acquisition Unit 112: Information Management Department 113: Data Generation Department 114: Model generation unit 115: Information estimation unit 116: Information output unit 130: Information storage area 1000: Water treatment system DT0: Time information DT1: Weather information DT1a: Weather information DT2: Water amount information DT2a: Water amount information DT2b: Water volume information DT11: Training data DT12: Training data MD1: Learning model MD2: Learning model P1: Pump P2: Pump P3: Pump T:Storage tank

Claims

1. A first learning model is generated by performing machine learning on multiple first training data sets, each corresponding to a plurality of first time periods, which each include first weather information indicating the weather conditions in the first region for each time period, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time periods corresponding to each time period. A water demand forecasting program that causes a computer to execute a process that stores the generated first learning model in a memory unit.

2. The water demand forecasting program according to claim 1, wherein the first weather information includes at least one of the following: information indicating the weather in the first region for each time period; information indicating the temperature in the first region for each time period; and information indicating the amount of precipitation in the first region for each time period.

3. Furthermore, a second learning model is generated by performing machine learning on a plurality of second training data sets corresponding to each of the plurality of first time periods, which do not include the first weather information corresponding to each time period, but each includes the time-series data of the first water volume information corresponding to each time period. The water demand forecasting program according to claim 2, which causes a computer to perform the process of storing the generated second learning model in a memory unit.

4. A first learning model is generated by machine learning a plurality of first training data, each corresponding to a plurality of first time zones, which each includes first weather information indicating the weather conditions in the first region for each time zone, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time zones corresponding to each time zone. Second weather information indicating the weather conditions in the second region for the second time zone is input to the first learning model, and time series data of second water volume information indicating the amount of treated water needed in the second region for one or more time zones corresponding to the second time zone, which is output in conjunction with the input of the second weather information to the first learning model, is acquired. A water demand forecasting program that causes a computer to perform processing to output time-series data of the acquired second water volume information.

5. In the process of acquiring the time-series data of the second water volume information, If the second weather information satisfies predetermined conditions, the second weather information is input to the first learning model. If the second weather information does not satisfy the predetermined conditions, the second learning model is generated by machine learning of the plurality of second training data corresponding to each of the plurality of first time periods, which do not include the first weather information corresponding to each time period, but each includes the first water volume information corresponding to each time period. The second learning model is then input with respect to the input of the second time period information to the second learning model, and time-series data of other second water volume information indicating the amount of treated water required in the second region for one or more time periods corresponding to the second time period is obtained. The water demand forecasting program according to claim 4, wherein the process for outputting time-series data of the second water volume information outputs the acquired time-series data of the second water volume information or time-series data of the other second water volume information.

6. A model generation unit generates a first learning model by performing machine learning on multiple first training data sets, each of which corresponds to multiple first time periods, and each first training data set includes first weather information indicating the weather conditions in the first region for each time period, and first water volume information time series data indicating the amount of treated water required in the first region for one or more time periods corresponding to each time period. An information processing device having a storage unit for storing the generated first learning model.

7. An information estimation unit inputs second meteorological information indicating the weather conditions in a second region of a second time zone to a first learning model generated by machine learning on a plurality of first training data, each of which corresponds to a plurality of first time zones, and each training data includes first meteorological information indicating the weather conditions in a first region of a second time zone, and time-series data of first water volume information indicating the amount of treated water required in the second region of one or more time zones corresponding to each time zone. The information estimation unit then acquires time-series data of second water volume information indicating the amount of treated water required in the second region of one or more time zones corresponding to the second time zone, which is output in conjunction with the input of the second meteorological information to the first learning model. An information processing device having an information output unit that outputs time-series data of the acquired second water volume information.

8. A first learning model is generated by performing machine learning on multiple first training data sets, each corresponding to a plurality of first time periods, which each include first weather information indicating the weather conditions in the first region for each time period, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time periods corresponding to each time period. A water demand forecasting method in which a computer performs the process of storing the generated first learning model in a memory unit.

9. A first learning model is generated by machine learning a plurality of first training data, each corresponding to a plurality of first time zones, which each includes first weather information indicating the weather conditions in the first region for each time zone, and first water volume information time series data indicating the amount of treated water needed in the first region for one or more time zones corresponding to each time zone. Second weather information indicating the weather conditions in the second region for the second time zone is input to the first learning model, and time series data of second water volume information indicating the amount of treated water needed in the second region for one or more time zones corresponding to the second time zone, which is output in conjunction with the input of the second weather information to the first learning model, is acquired. A water demand forecasting method in which a computer performs processing to output time-series data of the acquired second water volume information.

Images