Water level estimation method, water level estimation model generation method, water level estimation apparatus, water level estimation model generation apparatus, and program

Abstract

To predict a water level of a channel of a sewer while sufficiently taking reliability of prediction into consideration.SOLUTION: A water level estimation method executed by a computer includes a step of estimating estimated water level information by inputting estimation rain amount information relating to measured values of rain amounts along a time series into a plurality of water level estimation models that have been learned using teacher data including teacher rain amount information relating to measured values of rain amounts along a time series and teacher water level information relating to measured values of water levels of a sewer channel along a time series, and a step of presenting an estimation result of the estimating step to a user.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a water level estimation method, a water level estimation model generation method, a water level estimation device, a water level estimation model generation device, and a program. [Background technology]

[0002] In general, sewer channels (e.g., open channels) are at increased risk of flooding due to rain. Avoiding overflow is an important aspect of sewer management.

[0003] For example, Patent Document 1 discloses a learning device and a sewage inflow volume prediction device that enables accurate prediction of the amount of sewage inflow into a sewage treatment plant for the next day, allowing operators to carry out appropriate treatment. This learning device includes an actual measurement data acquisition unit that acquires actual measurement data on the amount of sewage inflow into a sewage treatment plant, a precipitation data acquisition unit that acquires precipitation data, and a prediction model construction unit that uses deep learning to construct a sewage inflow prediction model that predicts the next day's sewage inflow from the actual measurement data on the amount of sewage inflow and the precipitation data. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2020-170339 Summary of the Invention [Problem to be solved by the invention]

[0005] The water level in the sewer canal rises in a short period of time. Therefore, compared to other waterways, water level prediction errors are more likely to be large. Therefore, when predicting the water level of sewerage channels, it is important to take prediction error into account.

[0006] The sewage inflow prediction model of Patent Document 1 predicts water levels using a single sewage inflow prediction model, and therefore does not fully consider the reliability of the prediction.

[0007] An object of the present invention is to predict the water level in a sewer channel while taking into full consideration the reliability of the prediction. [Means for solving the problem]

[0008] One aspect of the present invention is 1. A computer-implemented water level estimation method, comprising: The method comprises a step of estimating estimated water level information by inputting estimation rainfall information relating to measured values ​​of rainfall over time into a plurality of water level estimation models trained using training data including training rainfall information relating to measured values ​​of rainfall over time and training water level information relating to measured values ​​of sewer water levels over time, a step of presenting an estimation result of the estimating step to a user; This is a method for estimating water levels. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing a configuration of an information processing system according to an embodiment of the present invention; [Figure 2] FIG. 2 is a functional block diagram of the information processing system of FIG. 1. [Figure 3] FIG. 1 is an explanatory diagram of an overview of the present embodiment. [Figure 4] FIG. 2 is a diagram showing a data structure of measurement data according to the present embodiment. [Figure 5] FIG. 2 is a diagram showing the overall flow of information processing according to the present embodiment. [Figure 6] 10 is a flowchart of a learning process according to the present embodiment. [Figure 7] FIG. 7 is an explanatory diagram of the processing of FIG. 6. [Figure 8] FIG. 2 is an explanatory diagram of training data (rainfall information) according to the present embodiment. [Figure 9] FIG. 2 is an explanatory diagram of teacher data (water level information) according to the present embodiment. [Figure 10] 10 is a flowchart of an estimation process according to the present embodiment. [Figure 11] FIG. 11 is an explanatory diagram of the estimation in FIG. 10. [Figure 12] FIG. 11 is an explanatory diagram of the estimation in FIG. 10. [Figure 12] FIG. 10 is a sequence diagram of the process of presenting an estimated water level in this embodiment. [Figure 13] 13 is a diagram showing an example of a screen displayed in the information processing of FIG. 12. FIG. [Figure 14] FIG. 10 is an explanatory diagram of an overview of Modification 1. [Figure 15] FIG. 10 is a diagram showing the data structure of measurement data in Modification 1. [Figure 16] FIG. 10 is a diagram showing the data structure of a database according to Modification 1. [Figure 17] 10 is a flowchart of a learning process according to the first modification. [Figure 18] 10 is a flowchart of a process for estimating a water level in Modification 1. [Figure 19] FIG. 10 is a sequence diagram of the process of presenting an estimated water level in Modification 1. [Figure 20] FIG. 20 is a diagram showing an example of a screen displayed in the processing of FIG. 19. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. In the drawings for explaining the embodiment, the same components are generally designated by the same reference numerals, and repeated description thereof will be omitted.

[0011] (1) Information processing system configuration The configuration of the information processing system will be described below. Fig. 1 is a block diagram showing the configuration of the information processing system of this embodiment. Fig. 2 is a functional block diagram of the information processing system of Fig. 1.

[0012] As shown in FIG. 1, the information processing system 1 includes a client device 10, a server 30, and an external server WS. The server 30 is connected to the client device 10 and the external server WS via a network (for example, the Internet) NW.

[0013] The client device 10 is a computer (an example of an "information processing device") that transmits and receives information to and from the server 30. The client device 10 is, for example, a smartphone, a tablet terminal, or a personal computer.

[0014] The server 30 is a computer (an example of an "information processing device") that transmits and receives information to and from the client device 10. The server 30 is, for example, a web server.

[0015] The external server WS provides meteorological information (for example, measured values ​​of rainfall information and measured values ​​of sewer water level information).

[0016] (1-1) Client device configuration The configuration of the client device 10 will be described.

[0017] As shown in FIG. 2, the client device 10 includes a storage device 11, a processor 12, an input / output interface 13, and a communication interface 14.

[0018] The storage device 11 is configured to store programs and data, and is, for example, a combination of a read-only memory (ROM), a random access memory (RAM), and a storage (for example, a flash memory or a hard disk).

[0019] The programs include, for example, the following programs: OS (Operating System) programs Parameters, functions, and models used in information processing Applications that process information (e.g., web browsers)

[0020] The data includes, for example, the following data: Databases referenced in information processing Data obtained by performing information processing (i.e., the results of performing information processing)

[0021] The processor 12 is configured to realize the functions of the client device 10 by running a program stored in the storage device 11. The processor 12 is, for example, a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or a combination thereof.

[0022] The input / output interface 13 is configured to receive user instructions from an input device connected to the client device 10 and to output information to an output device connected to the client device 10 . The input device is, for example, a keyboard, a pointing device, a touch panel, or a combination thereof. The output device is, for example, a display.

[0023] The communication interface 14 is configured to control communication between the client device 10 and the server 30 .

[0024] (1-2) Server configuration The configuration of the server 30 will be described.

[0025] As shown in FIG. 2, the server 30 includes a storage device 31, a processor 32, an input / output interface 33, and a communication interface .

[0026] The storage device 31 is configured to store programs and data, and is, for example, a combination of ROM, RAM, and storage (for example, flash memory or a hard disk).

[0027] The programs include, for example, the following programs: OS programs Parameters, functions, and models used in information processing Application programs that perform information processing

[0028] The data includes, for example, the following data: Databases referenced in information processing - Results of information processing

[0029] The processor 32 is configured to implement the functions of the server 30 by running a program stored in the storage device 31. The processor 32 is, for example, a CPU, a GPU, an ASIC, an FPGA, or a combination thereof.

[0030] The input / output interface 33 is configured to receive user instructions from an input device connected to the server 30 and to output information to an output device connected to the server 30 . The input device is, for example, a keyboard, a pointing device, a touch panel, or a combination thereof. The output device is, for example, a display.

[0031] The communication interface 34 is configured to control communication between the server 30 and the client device 10 and the external server WS.

[0032] (2) Overview of the embodiment An outline of this embodiment will be explained below with reference to Fig. 3, which is an explanatory diagram of the outline of this embodiment.

[0033] As shown in FIG. 3, the server 30 acquires rainfall information for estimation from an external server WS.

[0034] A water level estimation model is stored in the server 30. The water level estimation model is a trained model based on a neural network. The server 30 inputs the rainfall information for estimation into the water level estimation model. The water level estimation model outputs estimated water level information based on the rainfall information for estimation. The estimated water level information is information that indicates the predicted water level of the sewer channel estimated from the rainfall information for estimation. The sewer is, for example, at least one of the following: ·Drainage ·culvert

[0035] When a user accesses the server 30 using the client device 10, the server 30 transmits estimated water level information to the client device 10.

[0036] This allows the user to know the predicted results of the water level in the sewer channel based on the rainfall information for estimation.

[0037] (3) Data structure of measurement data The data structure of the measurement data in this embodiment will be described. FIG. 4 is a diagram showing the data structure of the measurement data in this embodiment.

[0038] As shown in FIG. 4, the measurement data includes measured rainfall information and measured water level information.

[0039] The measured rainfall information is information relating to measured values ​​of rainfall over time. The measured rainfall information includes a timestamp and rainfall information. The timestamp is information about the date and time when the rainfall was measured. The rainfall information is a measured value of the amount of rainfall.

[0040] The measured water level information is information relating to the measured values ​​of the water level in the sewer channel over time. The measured water level information includes a timestamp and water level information. The timestamp is information about the date and time when the water level was measured. The water level information is a measured value of the water level.

[0041] (4) Information processing The information processing of this embodiment will be described below with reference to Fig. 5, which is a diagram showing the overall flow of the information processing of this embodiment.

[0042] As shown in FIG. 5, the information processing of this embodiment includes learning (OP1), estimating the water level (OP2), and presenting the estimated water level (OP3).

[0043] Learning (OP1) is a process for generating a water level estimation model.

[0044] Water level estimation (OP2) is a process of estimating the water level using a water level estimation model.

[0045] Presentation of estimated water level (OP3) is a process of presenting the results of water level estimation (OP2) to the user.

[0046] (4-1) Learning Learning (OP1) of this embodiment will be described. Fig. 6 is a flowchart of the learning process of this embodiment. Fig. 7 is an explanatory diagram of the process of Fig. 6. Fig. 8 is an explanatory diagram of training data (rainfall information) of this embodiment. Fig. 9 is an explanatory diagram of training data (water level information) of this embodiment.

[0047] The process of FIG. 6 is triggered when an administrator of the server 30 (for example, a provider of estimated water level information) gives a predetermined instruction to the server 30 (for example, an instruction to start learning).

[0048] [Acquisition of measurement data (S1130)] As shown in FIG. 6, the server 30 executes acquisition of measurement data. Specifically, the external server WS stores measurement data. The measurement data includes measured rainfall information (FIG. 4) and measured water level information (FIG. 4) (FIG. 7). The processor 32 acquires the measurement data from the external server WS.

[0049] [Generating training data (S1131)] The server 30 generates the training data. Specifically, the processor 32 randomly extracts a portion of the measured rainfall information from the measurement data obtained in step S1130 by applying bootstrap sampling to the measurement data. The processor 32 randomly extracts a portion of the measured water level information from the measurement data obtained in step S1130 by applying bootstrap sampling to the measurement data. The processor 32 stores the extracted combination of measured rainfall information and extracted measured water level information in the storage device 31 as training data Ti (i is a natural number from 1 to N (N≧2)).

[0050] The server 30 repeats step S1131 N times to generate N pieces of training data T1 to TN. That is, the server 30 generates a plurality of pieces of training data T1 to TN from one piece of measurement data.

[0051] As shown in Figures 8 and 9, the teacher's rainfall information i and teacher's water level information i of certain teacher data Ti are different from the teacher's rainfall information j and teacher's water level information j of other teacher data Tj (j is a natural number from 1 to N). That is, at least a part of the teacher rainfall information i and teacher water level information i of a certain teacher data Ti is information that is not included in other teacher data Tj. On the other hand, at least a part of the teacher rainfall information i and teacher water level information i of a certain teacher data Ti may be included in other teacher data Tj (that is, between a certain teacher data Ti and other teacher data Tj, the teacher rainfall information i and teacher water level information i may overlap with some of the teacher rainfall information j and teacher water level information j).

[0052] [Generate model set (S1132)] After step S1131, the server 30 executes generation of a model set. Specifically, the processor 32 generates a water level estimation model Mi by applying deep learning to the training data Ti obtained in step S1131, and stores the water level estimation model Mi in the storage device 31. The water level estimation model Mi is a model that defines the correlation between rainfall and water level. Deep learning is, for example, at least one of the following: Deep Neural Network (DNN) Recurrent Neural Network (RNN) ·Long Short Term Memory (LSTM)

[0053] The server 30 generates a plurality of water level estimation models M1 to MN by repeating step S1132 N times. The water level estimation models M1 to MN are different from each other. That is, the server 30 generates a plurality of water level estimation models M1 to MN from one measurement data.

[0054] (4-2) Estimation of water level The estimation of water level (OP2) of this embodiment will be described. Fig. 10 is a flowchart of the estimation process of this embodiment. Fig. 11 is an explanatory diagram of the estimation in Fig. 10. Fig. 12 is an explanatory diagram of the estimation in Fig. 10.

[0055] The trigger in FIG. 10 is that a certain period of time has elapsed since the previous execution of the process in FIG. 10 was completed.

[0056] [Acquisition of rainfall information for estimation (S1230)] As shown in FIG. 10, the server 30 executes acquisition of rainfall information for estimation. Specifically, the processor 32 acquires rainfall information for estimation from the external server WS. The rainfall information for estimation is rainfall information that is not used in learning (OP1). Rainfall information not used in learning (OP1) is classified into information older than the learning rainfall information and information newer than the learning rainfall information. The information older than the learning rainfall information is rainfall information that precedes the oldest rainfall information among the learning rainfall information. The information newer than the learning rainfall information includes at least one of rainfall information subsequent to the most recent rainfall information in the learning rainfall information and rainfall information older than the time step S1230 was performed. For example, in learning (OP1), if rainfall information from January 1, 2024 to March 31, 2024 is used as learning rainfall information and step S1230 is executed on May 1, 2024, the estimation rainfall information includes at least one of rainfall information before December 31, 2023 and rainfall information from April 1, 2024 to April 30, 2024.

[0057] [Inference (S1231)] After step S1230, the server 30 executes inference. Specifically, when the processor 32 inputs the estimation rainfall information obtained in step S1230 into the water level estimation models M1 to MN included in the model set stored in the memory device 31, each water level estimation model M1 to MN outputs an inference result E1 to EN according to the estimation rainfall information.

[0058] [Calculation of estimated water level and confidence interval (S1232)] After step S1231, the server 30 performs calculation of the estimated water level and the confidence interval.

[0059] In the first example of step S1232, the processor 32 calculates a first average value μ1i and a first standard deviation σ1i based on the inference result Ei obtained in step S1231. The processor 32 stores the first average value μ1i as estimated water level information in the storage device 31. The first average value μ1i is the average value of the N inference results E1 to EN. The processor 32 stores the first standard deviation σ1i as confidence interval information in association with the estimated water level information (μ1i). The first standard deviation σ1i is the standard deviation of the N inference results E1 to EN.

[0060] In the second example of step S1232, following the first example, the processor 32 calculates the second mean value μ2i and the second standard deviation σ2i by applying the first mean value μ1i and the first standard deviation σ1i to a predetermined calculation model (e.g., a stochastic differential equation). The second average value μ2i is the time average of the first average value μ1i. The second standard deviation σ2i is the standard deviation of the first standard deviation σ1i.

[0061] For example, the stochastic differential equation is expressed as Equation 1. dSt = St (μdt + σdWt) …(Equation 1) ·St: Estimated water level at time t μdt: Estimated average rate of increase in water level per unit time ·σdWt…Error in estimated water level

[0062] The processor 32 stores the second average value μ2i in the storage device 31 as estimated water level information. The processor 32 stores the second standard deviation σ2i as confidence interval information in association with the estimated water level information (μ2i).

[0063] The estimated water level information indicates, for example, the water level t minutes (for example, t=30, 60, and 90) after step S1233 is executed. The reliability interval information indicates an interval of the estimated water level information where the reliability of the estimation is at or above a certain level.

[0064] [Generate warning message (S1233)] After step S1232, the server 30 generates a warning message. Specifically, when the estimated water level information obtained in step S1232 satisfies the warning condition, the processor 32 generates a warning message and stores the warning message in the storage device 31. The warning message is, for example, a message that alerts the user to the risk of an accident caused by the high water level. The warning condition is, for example, at least one of the following: - When the water level indicated by the estimated water level information is equal to or greater than the specified warning threshold. - When the water level indicated by the estimated water level information remains above the specified warning threshold for a specified period of time or longer. - When the rate of rise per unit time of the water level indicated by the estimated water level information is equal to or greater than the warning threshold. - When the water level indicated by the estimated water level information is equal to or exceeds the warning threshold after a certain period of time (for example, 10 minutes) has elapsed

[0065] Step S1233 can be omitted if the estimated water level information does not satisfy the warning conditions.

[0066] (4-3) Presentation of estimated water levels The presentation of the estimated water level (OP3) of this embodiment will be described. Fig. 12 is a sequence diagram of the process of presenting the estimated water level of this embodiment. Fig. 13 is a diagram showing an example of a screen displayed in the information processing of Fig. 12.

[0067] The trigger in FIG. 12 is an instruction by the user (for example, an instruction to access a predetermined website or an instruction to start a water level prediction application installed on the client device 10).

[0068] [Presentation request (S1310)] As shown in FIG. 12, the client device 10 executes a presentation request. Specifically, the processor 12 displays a screen P1310 (FIG. 13) on the display.

[0069] The screen P1310 includes an operation object B1310.

[0070] The operation object B 1310 is an object that receives a user instruction to display estimated water level information.

[0071] When the user operates the operation object B 1310, the client device 10 transmits presentation request data to the server 30.

[0072] [Selection of estimated water level information (S1330)] After step S1310, the server 30 executes the selection of estimated water level information. Specifically, the processor 32 selects the estimated water level information stored in the storage device 31.

[0073] [Proposal Response (S1331)] After step S1330, the server 30 executes the presentation response. Specifically, the processor 32 transmits the presented response data to the client device 10. The presented response data includes, for example, the following information: Estimated water level information selected in step S1330

[0074] [Presentation of Estimation Results (S1311)] After step S1332, the client device 10 executes presentation of the estimation result. Specifically, the processor 12 displays screen P1311 (FIG. 13) on the display.

[0075] The screen P1311 includes a display object A1311, an operation object B1311, and image objects IMG13110 to IMG13111.

[0076] The display object A1311 is an object that displays a warning message included in the estimated response data.

[0077] The operation object B 1311 is an object that receives a user instruction to switch the image object IMG 13110.

[0078] Image object IMG13110 is an image of a graph corresponding to estimated water level information. The horizontal axis of the graph is time. The vertical axis of the graph is the estimated water level. The image object IMG13111 is an image showing the confidence interval corresponding to the confidence interval information added to the estimated water level information.

[0079] FIG. 13 shows an example in which the image object IMG13110 is an image of a graph of estimated water level information at t=30. When the user operates the operation object B 1311 to give an instruction to change the value of t, the image object IMG 13110 switches to an image of a graph of estimated water level information for t=60 or 90.

[0080] (5) Summary of this embodiment According to this embodiment, the estimated water level information is estimated using a plurality of water level estimation models. This makes it possible to predict the water level in sewer channels while fully considering the reliability of the prediction.

[0081] According to this embodiment, the average value of the outputs of the water level estimation models corresponding to the rainfall information for estimation may be calculated as the estimated water level information. This allows the water level in sewer channels to be predicted while taking into consideration the reliability of the prediction.

[0082] According to this embodiment, the estimated water level information and confidence interval information relating to the confidence interval of the estimated water level information may be estimated. This allows the reliability of the prediction to be presented to the user.

[0083] According to this embodiment, the standard deviation of the output of each water level estimation model according to the rainfall information for estimation may be calculated as the confidence interval information. This allows the reliability of the prediction to be presented to the user.

[0084] According to this embodiment, the estimating step may calculate a second average value as estimated water level information by providing a first average value of the output of each water level estimation model to a predetermined stochastic differential model, and calculate a second standard deviation as confidence interval information by providing a first standard deviation of the output of each water level estimation model to a calculation model. This can improve the accuracy of the water level estimation information and the confidence interval.

[0085] (6) Variations A modification of this embodiment will now be described.

[0086] (6-1) Variation 1 A description will be given of Modification 1. Modification 1 is an example in which water level information is estimated using a model set for each region. (6-1-1) Overview of Modification 1 An outline of Modification 1 will be described below. Fig. 14 is an explanatory diagram of an outline of Modification 1.

[0087] As shown in FIG. 14, the server 30 acquires rainfall information for estimation from the external server WS, similarly to FIG.

[0088] A plurality of model sets MSa to MSb are stored in the server 30. Each model set includes a water level estimation model. Each model set is associated with area information. The server 30 selects a model set (for example, model set MSb) associated with the area information specified by the user (for example, area information indicating area B). The server 30 inputs the rainfall information for estimation associated with the area information (for example, area information indicating area B) to the water level estimation model included in the selected model set. The water level estimation model outputs estimated water level information based on the rainfall information for estimation.

[0089] When a user accesses the server 30 using the client device 10, the server 30 transmits estimated water level information to the client device 10.

[0090] This allows the user to know the water level prediction results for a desired area based on the rainfall information for estimation.

[0091] (6-1-2) Data structure of variant 1 The data structure of the first modification will be described.

[0092] (6-1-2-1) Data structure of measurement data in Modification 1 The data structure of the measurement data in the first modification will be described. FIG. 15 is a diagram showing the data structure of the measurement data in the first modification.

[0093] As shown in FIG. 15, the measurement data is associated with area information. The measurement data associated with the area information indicating the area A includes measured rainfall information associated with the area information and measured water level information associated with the area information. The measurement data associated with the area information indicating area B includes measured rainfall information associated with the area information and measured water level information associated with the area information.

[0094] The measured rainfall information and measured water level information are the same as those in FIG.

[0095] (6-1-2-2) Database of Modification 1 The database of the first modification will be described. FIG. 16 is a diagram showing the data structure of the database of the first modification.

[0096] (6-1-2-2-1) Model set database of variant 1 The model set database of the first modification will be described.

[0097] As shown in Fig. 16, the model set database stores model set information. The model set information is information related to a model set. A model set is a combination of multiple water level estimation models. The model set database includes a "model set ID" field, a "model set name" field, a "model set" field, and a "region" field. Each field is associated with the others.

[0098] The "model set ID" field stores model set identification information, which is information for identifying a model set.

[0099] The "model set name" field stores model set name information, which is information about the name of the model set.

[0100] The "model set" field stores the model set.

[0101] The "Region" field stores region information associated with the measurement data used to generate the model set.

[0102] (6-1-2-2-2) Estimation result database of variant 1 The estimation result database of the first modification will be described.

[0103] As shown in FIG. 16, the estimation result database stores estimated water level information. The estimation result database includes an "estimation result ID" field, an "estimated water level" field, a "warning message" field, and a "region" field. Each field is associated with the others.

[0104] The "Estimation Result ID" field stores estimation result identification information. The estimation result identification information is information that identifies the estimation result (i.e., estimated water level information).

[0105] The "Estimated Water Level" field stores estimated water level information.

[0106] In the "Warning message" field, a warning message is generated.

[0107] The "Area" field stores area information associated with the model set used to generate the estimated water level information. The region information associated with the model set indicates, for example, at least one of the following regions: Areas determined by the location of sewerage systems Areas determined by the local government that manages the sewer system ·Sewerage drainage area

[0108] (6-1-3) Information processing of variation 1 The information processing of the first modification will be described.

[0109] (6-1-3-1) Study of Modification 1 The learning (OP1) of Modification 1 will be described below. Fig. 17 is a flowchart of the learning process of Modification 1.

[0110] The trigger for the process in FIG. 17 is the same as in FIG.

[0111] [Selection of measurement data by region (S2130)] As shown in FIG. 17, the server 30 selects measurement data for each region. Specifically, the external server WS stores measurement data for each region (i.e., measured rainfall information for each region and measured water level information for each region). The measurement data is stored in association with region information. The processor 32 acquires the measurement data for each region from the external server WS.

[0112] [Generating training data (S1131) ~ Generating model sets (S1132)] After step S2130, the server 30 executes the processes from generating training data to generating a model set, similar to FIG.

[0113] [Database Update (S2131)] After step S1132, the server 30 updates the database. Specifically, processor 32 adds a new record to the model set database. The following information is stored in each field of the new record: "Model Set ID" field: New model set identification information "Model set" field: A model set consisting of multiple water level estimation models M1 to MN obtained in step S1132 "Region" field: Region information associated with the measurement data used in step S1131

[0114] (6-1-3-2) Estimation of water level for variant 1 The following describes the water level estimation (OP2) of Modification 1. Fig. 18 is a flowchart of the process of estimating the water level of Modification 1.

[0115] The trigger for the process in FIG. 18 is the same as in FIG.

[0116] [Acquisition of regional rainfall information for estimation (S2230)] As shown in FIG. 18, the server 30 executes acquisition of rainfall information for estimation for each region. Specifically, the external server WS stores rainfall information for estimation for each region, which is associated with region information. The processor 32 acquires rainfall information for estimation for each region from the external server WS.

[0117] [Model Set Selection (S2231)] After step S2230, the server 30 executes the selection of a model set. Specifically, the model set database (FIG. 16) is referenced to select a model set associated with the area information associated with the rainfall information for estimation obtained in step S2230.

[0118] [Inference (S2232)] After step S2231, the server 30 executes inference. Specifically, when processor 32 inputs the estimation rainfall information obtained in step S1230 to water level estimation models M1 to MN included in the model set selected in step S2231, each water level estimation model M1 to MN outputs inference results E1 to EN according to the estimation rainfall information.

[0119] [Calculation of estimated water level and confidence interval (S1232)] After step S2232, the server 30 executes the calculation of the mean and standard deviation to the calculation of the confidence interval, similarly to FIG.

[0120] [Generate warning message (S2233)] After step S1232, the server 30 generates a warning message. Specifically, when the estimated water level information obtained in step S1232 satisfies the warning condition, the processor 32 generates a warning message and stores it in the storage device 31. The warning message is, for example, a message that alerts the user to the risk of an accident caused by the high water level. The warning condition is, for example, at least one of the following: If the water level indicated by the estimated water level information is equal to or greater than the warning threshold associated with the area information associated with the model set selected in step S2231 If the water level indicated by the estimated water level information is equal to or greater than the warning threshold associated with the area information associated with the model set selected in step S2231 for a predetermined period of time or longer. If the rate of rise per unit time of the water level indicated by the estimated water level information is equal to or greater than the warning threshold associated with the area information associated with the model set selected in step S2231 If the water level indicated by the estimated water level information after a predetermined time (for example, 10 minutes) has elapsed is equal to or greater than the warning threshold associated with the area information associated with the model set selected in step S2231.

[0121] Step S2233 can be omitted if the estimated water level information does not satisfy the warning conditions.

[0122] [Database Update (S2234)] After step S2233, the server 30 updates the database. Specifically, the processor 32 adds a new record to the estimation result database (FIG. 16). The following information is stored in each field of the new record: "Estimation result ID" field: New estimation result identification information "Region" field: Region information associated with the model set selected in step S2231 "Estimated water level" field: Estimated water level information obtained in step S1232 "Warning Message" field: Warning message obtained in step S1233

[0123] (6-1-3-3) Presentation of estimated water level for variant 1 The following describes the presentation of the estimated water level (OP3) in Modification 1. Fig. 19 is a sequence diagram of the process of presenting the estimated water level in Modification 1. Fig. 20 is a diagram showing an example of a screen displayed in the process in Fig. 19.

[0124] The trigger for the process in FIG. 19 is the same as in FIG.

[0125] [Presentation request (S2310)] As shown in FIG. 19, the client device 10 executes a presentation request. Specifically, the processor 12 displays a screen P2310 (FIG. 20) on the display.

[0126] The screen P2310 includes an operation object B1310 and a field object F2310. The operation object B 1310 is the same as that in FIG.

[0127] The field object F2310 is an object for receiving input of region information. The field object F2310 receives, for example, at least one of the following regional information: -Region information specified by the user The client device 10 includes a location information sensor (for example, a GPS (Global Positioning System) module) that acquires location information.

[0128] When the user inputs any region information into the field object F2310 and operates the operation object B1310, the client device 10 transmits presentation request data to the server 30. The presentation request data includes, for example, the following information: -Region information entered in field object F2310

[0129] [Selection of estimated water level information (S2330)] After step S2310, the server 30 executes the selection of estimated water level information. Specifically, the processor 32 refers to the estimation result database (FIG. 16) and selects estimated water level information associated with the area information included in the presentation request data.

[0130] [Proposal Response (S1331)] After step S2330, the server 30 executes the presentation response in the same manner as in FIG.

[0131] [Presentation of Estimation Results (S1311)] After step S1331, the client device 10 executes presentation of the estimation result in the same manner as in FIG.

[0132] (6-1-4) Summary of Modification 1 According to the first modification, the water level is estimated using a water level estimation model for each area designated by the user. This can further reduce the prediction error of the estimation.

[0133] (7) Other variations Other modifications will be described.

[0134] The storage device 11 may be connected to the client device 10 via a network NW. The storage device 31 may be connected to the server 30 via the network NW.

[0135] Each step of the above information processing can be executed by either the client device 10 or the server 30. For example, if the client device 10 is capable of executing all of the steps of the above-described information processing, the client device 10 functions as an information processing device that operates standalone without transmitting a request to the server 30 .

[0136] In this embodiment, a general-purpose water level estimation model may be generated by performing fine tuning using mutually similar water level estimation models from among a plurality of water level estimation models. By using a general-purpose water level estimation model, it is possible to predict the water level of sewerage channels without depending on the region or waterway, while fully considering the reliability of the prediction.

[0137] Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited to the above-described embodiments. Furthermore, the above-described embodiments can be improved or modified in various ways without departing from the spirit of the present invention. Furthermore, the above-described embodiments and modifications can be combined. [Explanation of symbols]

[0138] 1: Information processing system 10: Client device 11:Storage device 12: Processor 13: Input / output interface 14: Communication interface 30: Server 31: Storage device 32: Processor 33: Input / output interface 34: Communication interface

Claims

1. 1. A computer-implemented water level estimation method, comprising: The method comprises a step of estimating estimated water level information by inputting estimation rainfall information relating to measured values ​​of rainfall over time into a plurality of water level estimation models trained using training data including training rainfall information relating to measured values ​​of rainfall over time and training water level information relating to measured values ​​of sewer water levels over time, a step of presenting an estimation result of the estimating step to a user; Water level estimation method.

2. the estimating step calculates an average value of outputs of each water level estimation model according to the rainfall information for estimation as the estimated water level information; The water level estimation method according to claim 1 .

3. The estimating step estimates the estimated water level information and confidence interval information relating to a confidence interval of the estimated water level information. The water level estimation method according to claim 1 .

4. The estimating step calculates a standard deviation of an output of each water level estimation model according to the rainfall information for estimation as the confidence interval information. The water level estimation method according to claim 3.

5. The estimating step includes: calculating a second average value as the estimated water level information by applying a first average value of the output of each water level estimation model to a predetermined stochastic differential equation; a first standard deviation of the output of each water level estimation model is applied to the stochastic differential equation to calculate a second standard deviation as the confidence interval information; The water level estimation method according to claim 4.

6. receiving designation of area information from the user; Each water level estimation model is associated with local information. selecting a water level estimation model associated with area information designated by the user from the plurality of water level estimation models; the estimating step estimates estimated water level information for the area indicated by the area information by inputting rainfall information for estimation associated with the area information into the selected water level estimation model; A water level estimation method according to any one of claims 1 to 5.

7. The presenting step presents a warning message to the user when the estimated water level information satisfies a warning condition. A water level estimation method according to any one of claims 1 to 5.

8. receiving designation of area information from the user; The presenting step presents the warning message when the water level indicated by the estimated water level information satisfies a warning condition defined for each region indicated by the region information. The water level estimation method according to claim 7.

9. The teacher rainfall information is a part of rainfall information relating to rainfall along a time series, The estimation rainfall information is information of the rainfall information that is not included in the teacher rainfall information. A water level estimation method according to any one of claims 1 to 5.

10. A method for generating a water level estimation model using a computer, comprising: generating a plurality of pieces of teacher data by performing bootstrap sampling on measurement data including teacher rainfall information relating to time-series measurement values ​​of rainfall and teacher water level information relating to time-series measurement values ​​of sewer water levels; generating a plurality of water level estimation models by deep learning based on each training data; How to generate a water level estimation model.

11. The step of generating the plurality of teacher data uses part of the teacher rainfall information as teacher rainfall information, and storing, in a storage device, information that has not been used in the step of generating the plurality of pieces of teacher data from the teacher rainfall information, for use in inference using the water level estimation model. The method for generating a water level estimation model according to claim 10.

12. a means for estimating estimated water level information by inputting estimation rainfall information relating to measured values ​​of rainfall over time into a plurality of water level estimation models trained using training data including training rainfall information relating to measured values ​​of rainfall over time and training water level information relating to measured values ​​of sewer water levels over time; means for presenting an estimation result of the estimating means to a user; Water level estimation device.

13. Computer, inputting estimation rainfall information relating to the measured values ​​of rainfall over time into a plurality of water level estimation models trained using training data including training rainfall information relating to the measured values ​​of rainfall over time and training water level information relating to the measured values ​​of sewer water levels over time, thereby causing the models to function as a means for estimating estimated water level information; functioning as a means for presenting the estimation result of the estimating means to a user; Program for.

14. a means for generating a plurality of pieces of teacher data by performing bootstrap sampling on measurement data including teacher rainfall information relating to time-series measurement values ​​of rainfall and teacher water level information relating to time-series measurement values ​​of sewer water levels; A means for generating a plurality of water level estimation models by deep learning based on each training data is provided. Water level estimation model generator.

15. Computer, and functioning as a means for generating a plurality of pieces of teacher data by performing bootstrap sampling on measurement data including teacher rainfall information relating to measurement values ​​of rainfall over time and teacher water level information relating to measurement values ​​of sewer water levels over time; It serves as a means of generating multiple water level estimation models through deep learning based on each training data. Program for.

Images