JP2023169099A - Inflow prediction system - Google Patents

Abstract

To provide an inflow prediction system capable of accurately performing prediction of a waste water inflow amount in a pump station and the like.SOLUTION: An inflow prediction system 100 comprises: a facility data collection unit 113 for collecting facility data PD of a pump station 10 and the like; a weather data collection unit 110 for collecting weather data 185; a data accumulation unit 121 for accumulating the facility data PD and the weather data 185; a learning processing unit 150 for generating a learned model which predicts a waste water inflow amount on the basis of the facility data PD and the weather data 185; a prediction unit 160 for predicting a waste water inflow amount on the basis of the learned model, the facility data PD and the weather data 185; and a control unit 170 for controlling an operation state of the pump station 10 and the like by receiving a predicted value of the waste water inflow amount. The weather data 185 is weather data 185A including information on a weather value, a longitude, a latitude, and time, and is the weather data of a mesh part which overlaps a region of a processing section 183 of the pump station 10 and the like.SELECTED DRAWING: Figure 2

Description

The present invention relates to an inflow amount prediction system that predicts the amount of sewage flowing into a pump station, a pump building in a sewage treatment plant, a sewage treatment plant that does not have a pump building, or the like.

Conventionally, predicting the amount of sewage flowing into facilities such as pumping stations, pump buildings within sewage treatment plants, or sewage treatment plants without pump buildings has been a method of determining the quality of treated water based on the appropriate operating method of the facility. This is important in stabilizing water, planning appropriate personnel allocation for maintenance during rainy weather, and preventing flooding inside and outside the facility.

For this reason, conventionally, skilled engineers in charge of facility management would predict fluctuations in the amount of sewage inflow based on the characteristics of the facility, weather conditions, etc., and temporarily reduce the amount of sewage before a rainfall event occurs. Alternatively, the amount of sewage retained in sewage pipes, pumping stations, or sewage treatment plants can be reduced in advance by increasing the amount of treatment, or some of the inflow water that has increased during rainy days can be stored in the pipes and discharged after the rain event ends. We are taking measures such as processing.

On the other hand, in recent years, a method of predicting the amount of sewage inflow using machine learning techniques has been studied as a method that can appropriately predict the amount of sewage inflow without requiring the use of skilled engineers.

Patent Document 1 discloses that a sewage inflow prediction model and a rainwater inflow prediction model are separately constructed using rainwater inflow data and sewage inflow data that are generated separately from sewage inflow data to pump wells, and calendar data is Input the meteorological data into a sewage inflow prediction model using case-based reasoning to generate sewage inflow prediction data, input meteorological data into a rainwater inflow prediction model using a neural network to generate rainwater inflow prediction data, and generate sewage inflow prediction data. A sewage inflow amount prediction device is disclosed that generates sewage inflow amount prediction data by adding up amount prediction data and rainwater inflow amount prediction data.

Patent Document 2 discloses that a user input section specifies a prediction step width indicating the time length from the current time to a future time at which the amount of sewage inflow is to be predicted, using a time interval of a predetermined length as a unit. The data collection unit collects the measured value of the sewage inflow rate for each time interval, and the short-term data storage unit stores the measured value of the sewage inflow rate in each time interval over a time period longer than multiple times the prediction step width. The measured value is stored, and the predicted value calculation unit repeatedly uses a prediction algorithm that calculates the predicted value of the next step based on the predicted value and measured value of a certain step, and calculates the amount of sewage inflow for each predicted step width. Sewage inflow A quantity prediction device is disclosed.

Patent Document 3 discloses that cumulative rainfall data obtained by converting observed values from a radar rain gauge into mesh-shaped rainfall intensity is stored as input data for learning, and on the other hand, the amount of rainwater flowing into a pump facility is stored as output data for learning. Based on the input data of cumulative rainfall data and the learning output data of rainwater inflow, the neural network learns and stores weighting coefficients using the back propagation method, and uses these weighting coefficients to calculate the cumulative rainfall data on the day of rainfall. A neural network-applied rainwater inflow prediction device that predicts and outputs the rainwater inflow using a neural network based on a determination value is disclosed.

Non-Patent Document 1 discloses a technique for predicting the amount of sewage flowing into a sewage treatment plant using machine learning (LSTM and CNN).

Japanese Patent Application Publication No. 2006-002401 Japanese Patent Application Publication No. 2016-211877 Japanese Patent Application Publication No. 5-134715

“Prediction of sewage flow using AI/Water Agency Co., Ltd.” (Academic journal “EICA”, Vol. 26, No. 2/3, P. 35-38, 2021)

However, in the above-mentioned Patent Documents 1 and 2 and Non-Patent Document 1, a means for obtaining rainfall amount data used as an explanatory variable in predicting the amount of sewage inflow with high accuracy is not disclosed.

Further, in Patent Document 3, as a means for obtaining highly accurate rainfall data, mesh-shaped rainfall data is used, and rainfall data in a predetermined mesh range of the mesh of the rainfall data is set as the cumulative rainfall. However, if the mesh range includes a mesh range that does not contribute to the amount of inflow to a pump station, etc., the accuracy of predicting the amount of sewage inflow may decrease.

The present invention has been made in view of the above points, and its purpose is to provide an inflow amount prediction system that can accurately predict the amount of sewage inflow at a pump station, a pump building in a sewage treatment plant, a sewage treatment plant, etc. It is about providing.

The present invention provides an inflow amount prediction system for predicting the amount of sewage inflow at a pump station or a sewage treatment plant, a pump building in the pump station or a sewage treatment plant, a pipeline facility having an inflow amount measuring means, or a sewage treatment plant. Alternatively, a facility data collection unit that collects facility data of a pipeline facility or a sewage treatment plant having an inflow measurement means, a weather data collection unit that collects weather data, and a data storage unit that accumulates the facility data and weather data. , a model that predicts the amount of sewage flowing into a pumping station or a pump building in a sewage treatment plant, a pipeline facility having an inflow amount measuring means, or a sewage treatment plant based on the facility data and meteorological data received from the data storage unit; a learning processing unit that learns and creates a trained model; a trained model storage unit that stores the trained model created by the learning processing unit; and a trained model storage unit that stores the trained model created by the learning processing unit, and a prediction unit that predicts the amount of sewage inflow at a future point; and a prediction unit that receives the predicted value of the amount of sewage inflow from the prediction unit, and generates a pump building in the pump station or sewage treatment plant, or a pipeline facility or sewage system having an inflow amount measuring means. a control unit that controls the operating state of the treatment plant, and the meteorological data used in the forecasting unit to predict the amount of sewage inflow is meteorological data in mesh units having information on meteorological values, longitude/latitude, and time. The data is characterized in that it is meteorological data of a mesh portion that overlaps with the pump station, the pump building in the sewage treatment plant, the pipeline facility having inflow measurement means, or the treatment area of the sewage treatment plant.
In addition to sewage treatment plants that have pump buildings, there are also sewage treatment plants that do not have pump buildings, and the present invention also includes such sewage treatment plants.
The specific control contents of the operation status of a pump station, a pump building in a sewage treatment plant, a pipeline facility having an inflow measuring means, or a sewage treatment plant (hereinafter also referred to as "pumping station, etc.") by the control unit include, for example: This includes adjusting the opening degree of the gate at the pumping station, adjusting the number of operating pumps, and adjusting the rotation speed of the pumps.
In addition to a flowmeter, a water level meter may be used as a means for measuring the inflow of a pipe facility, since the flow rate can be determined by the water level of sewage flowing inside the pipe.
According to the present invention, among the meteorological data in mesh units, the meteorological data of the mesh portion that overlaps with the area of the treatment area of the pumping station, etc., that is, the meteorological data that has a high influence on the amount of sewage flowing into the pumping station, etc. Since only the input value is used as an input value for predicting the amount of sewage inflow, the accuracy of predicting the amount of sewage inflow can be improved.

In addition to the above characteristics, the present invention is characterized in that the meteorological value is data including one or more of the following: precipitation, weather, temperature, wind speed, wind direction, and atmospheric pressure.
The meteorological value that most influences the prediction of sewage inflow is precipitation, but other meteorological values may also affect the prediction of sewage inflow. Therefore, by adding meteorological values other than precipitation according to the topography and meteorological characteristics of the treatment area, it is possible to further improve the accuracy of predicting the amount of sewage inflow.

In addition to the above features, the present invention provides that the weather data of the mesh portion overlapping with the area of the processing area is meteorological data of the mesh portion of the area overlapping with the area of the processing area on geographic information possessed by a geographic information system. It is characterized by
By using the geographical information possessed by the geographical information system on the network, it is possible to easily and accurately extract only the mesh part (area) that overlaps the processing area from the weather data, improving the prediction accuracy of sewage inflow. can be done.

In addition to the above features, the present invention is characterized in that the learning processing unit and the prediction unit perform learning and prediction using a model using machine learning.

In addition to the above characteristics, the present invention uses an RNN for the machine learning method, and uses the pump building or inflow amount measuring means in the pump station or sewage treatment plant as a time step that is one of the hyperparameters of the RNN. The peak time of sewage inflow in the pipeline facility or sewage treatment plant that has the pump station or the pump building in the sewage treatment plant or the treatment area of the pipeline facility or sewage treatment plant that has inflow measurement means. It is characterized by using the difference from the peak time of precipitation and/or the fluctuation period of the sewage inflow at the pumping station or the pump building in the sewage treatment plant, the pipeline facility having an inflow measurement means, or the sewage treatment plant. There is.
RNN (Recurrent Neural Network) is a deep learning method suitable for predicting time-series data whose values change over time. It is suitable for prediction using meteorological data every predetermined time.
According to the present invention, by using the difference between the peak time of the sewage inflow amount and the peak time of the precipitation amount for the treatment area and/or the fluctuation period of the sewage inflow amount as a time step, the amount of rainfall changes depending on the amount of sewage inflow. Since learning and prediction values can be calculated in consideration of contributing time, it is possible to reduce calculation costs or the number of learning trials, and improve the accuracy of sewage inflow prediction.

In addition to the above characteristics, the present invention also provides that the amount of precipitation is measured at the pump station or the pump building in the sewage treatment plant, or the pipe facility having an inflow measurement means, or the area that overlaps with the treatment area of the sewage treatment plant. The amount of precipitation for each time is the sum of the precipitation data for each mesh unit at the time, and the peak time of the precipitation and the pumping station or pump building in the sewage treatment plant, or the pipe facility or sewage system that has inflow measurement means. Using the difference between the peak time of the sewage inflow at the treatment plant and/or the fluctuation period of the sewage inflow at the pumping station, the pump building in the sewage treatment plant, the pipeline facility having an inflow measurement means, or the sewage treatment plant. The method is characterized in that the time step is determined based on the time step.
According to the present invention, the amount of precipitation for each time can be accurately calculated by summing up the amount of precipitation data for each mesh at the same time in a portion that overlaps with the area of the processing area. This makes it possible to accurately calculate the peak time of precipitation and improve the accuracy of predicting the amount of sewage inflow.

In addition to the above-mentioned features, the present invention provides that when the target for predicting the amount of sewage inflow is the pump station or a pump building in a sewage treatment plant, the amount of sewage inflow is calculated from one or more water tanks that the pump station or pump building has. The amount of fluctuation per unit time in the amount of water held in the pump, the amount of water pumped per unit time, and the amount of sewage stored in the inflow pipe outside the facility connected to the upstream side of the pump station or pump building per unit time. On the other hand, if the target for predicting the amount of sewage inflow is the sewage treatment plant, the amount of sewage inflow is maintained in one or more water tanks owned by the sewage treatment plant. It is characterized in that it is calculated by adding up the amount of change in the amount of water per unit time and the amount of change per unit time in the amount of sewage stored in the inflow pipe outside the facility connected to the upstream side of the sewage treatment plant.
In the case of a pump station or a pump building in a sewage treatment plant, the water tank whose water level fluctuates according to changes in the amount of sewage inflow corresponds to the water tanks that constitute the inlet culvert, settling basin, pump well, etc. In the case of a sewage treatment plant that does not have a water tank, this applies to water tanks that constitute the initial sedimentation tank, reaction tank, final sedimentation tank, etc.
According to the present invention, the amount of sewage inflow can be calculated with high accuracy even if a flow meter for measuring the amount of sewage flowing into a pump station or the like is not installed.

In addition to the above features, the present invention provides that the amount of variation per unit time in the amount of water held in each tank is calculated by multiplying the cross-sectional area of each tank by the variation in water level per unit time. It is a feature.
Thereby, the amount of variation per unit time in the amount of water held in each tank can be easily and accurately calculated.

In addition to the above characteristics, the present invention provides that the amount of fluctuation per unit time of the storage amount in the out-of-facility inflow pipe is determined by the water level of the water tank to which the out-of-facility inflow pipe is connected, and the gradient and pipe diameter of the out-of-facility inflow pipe. It is characterized in that it is calculated by referring to.
This makes it possible to easily and accurately calculate the amount of variation per unit time in the amount of storage in the outflow pipe outside the facility.

Furthermore, according to the present invention, it is possible to predict with high accuracy the amount of sewage inflow not only to the pump station or the pump building in the sewage treatment plant but also to a pipeline facility equipped with a flow rate measuring means such as a flow meter or a water level meter.

In addition to the above characteristics, the present invention also provides an upstream pumping station that collects and flows sewage from an upstream treatment area located upstream of the treatment area, or an upstream pump station located upstream of the treatment area. A pipeline facility having an inflow measurement means for the side treatment area is connected, and the predicted value of the sewage inflow at the upstream pumping station, determined by the inflow prediction system, is used to calculate the amount of sewage at the upstream pumping station. Calculate a predicted value of pumped water amount, and/or calculate a predicted value of sewage inflow amount in a pipeline facility having an inflow amount measuring means in the upstream treatment area using the inflow amount prediction system, In addition to the learned model, the facility data, and the meteorological data, the prediction unit uses the following methods to predict the amount of sewage inflow by the prediction unit in a pump building in a field or a sewage treatment plant, a pipeline facility having an inflow amount measuring means, or a sewage treatment plant. The present invention is characterized in that a predicted value of the amount of pumped water at the upstream pumping station and/or a predicted value of the amount of sewage inflow in the pipeline facility of the upstream treatment area is used.
According to the present invention, when sewage from an upstream treatment area is collected and flowed into a downstream treatment area by an upstream pumping station, the predicted value of the pumping amount of the upstream pumping station and the upstream treatment area are If a pipeline facility with inflow measurement means is installed, the predicted value of the sewage inflow rate at the pipeline facility in the upstream treatment area is the predicted value of the sewage inflow rate to the downstream pumping station, etc. Since it can be used as calculation data, it becomes possible to easily and accurately predict the amount of sewage inflow at the downstream pump station, etc.

According to the present invention, it is possible to accurately predict the amount of sewage inflow at a pump station, a pump building in a sewage treatment plant, a pipeline facility having an inflow amount measuring means, or a sewage treatment plant.

1 is a schematic configuration diagram showing an example of a pump station 10. FIG. 1 is a system configuration diagram showing an example of an inflow prediction system 100. FIG. It is a figure showing an outline of a processing procedure of a method of predicting the amount of sewage inflow. It is a figure showing the calculation procedure of meteorological data. It is a schematic configuration diagram showing an example of a sewage treatment plant 200 with a pump building. It is a schematic configuration diagram showing an example of a sewage treatment plant 300 that does not have a pump building. It is a schematic plan view at the time of connecting the upstream side pumping station 500 of the upstream side treatment area 501 to the upstream side of the treatment area 401 of the sewage treatment plant 400. It is a figure which shows the outline of the processing procedure of the sewage inflow amount prediction method in both the sewage treatment plant 400 and the upstream pump station 500.

Embodiments of the present invention will be described in detail below with reference to the drawings.
FIG. 1 is a schematic configuration diagram showing an example of a pumping station 10 controlled by an inflow prediction system 100 according to an embodiment of the present invention, and FIG. 2 is a system configuration diagram showing an example of the inflow prediction system 100.

The pump station 10 is installed to collect sewage discharged from homes, factories, etc. through a sewage pipe and send it to a sewage treatment plant, or to collect rainwater through a sewage pipe and discharge it into a river or the like. In the case of a combined sewer system, it is installed to combine sewage and rainwater and send it to a sewage treatment plant.

As shown in FIG. 1, for example, the pump station 10 includes an inlet culvert 13 consisting of a water tank into which sewage flows from an inflow pipe 11 outside the facility, and a water tank in which sewage is introduced from the inlet culvert 13 and solids such as garbage and sand are precipitated. a pump well 17 consisting of a water tank for introducing sewage from which solids have been removed in the settling basin 15, pumping it up with a pump P, and discharging it further downstream to a pipe (sewage pipe) 17A; and a pumping station. The inflow amount prediction system 100 predicts the amount of sewage flowing into the pump station 10 and controls the operating state of the pump station 10.

The upstream side of the out-of-facility inflow pipe 11 is connected to a sewage pipe line extending around a basin (treatment area) that collects sewage to the pumping station 10, and the sewage in the treatment area is collected into the out-of-facility inflow pipe 11. As mentioned above, the collected sewage may be only sewage, only rainwater, or a combination of sewage and rainwater.

Water level gauges 19 , 21 , and 23 are installed in the inflow channel 13 , the settling basin 15 , and the pump well 17 , respectively, to detect changes in the water level of each water tank and transmit each detected value to the inflow prediction system 100 . Note that the water level gauge 21 may not be installed in the settling basin 15, but in such a case, the water level in the settling basin 15 may be treated as being the same as the water level in the pump well 17.

A gate 25 is installed between the inflow channel 13 and the settling basin 15. The opening degree of this gate 25 is controlled by a control signal from the inflow prediction system 100.

A plurality of pumps P are installed in parallel (only one pump is shown in FIG. 1). The operating state (operation/stop and rotation speed) of each pump P is controlled by a control signal from the inflow prediction system 100. Pump 10 pumps up the sewage in pump well 17 and transfers it to the next stage via piping 17A.

As shown in FIG. 2, the inflow prediction system 100 includes a data collection unit 110 having a facility data collection unit 113 and a weather data collection unit 115, a storage unit 120 that stores various data, and a learning unit that creates a trained model. The system includes a processing section 150, a prediction section 160 that predicts the amount of sewage flowing into the pumping station 10, and a control section 170 that controls the operating state of the pumping station 10 based on the predicted value of the amount of sewage flowing into the pumping station 10. ing.

The facility data PD collected by the facility data collection unit 113 is data obtained from the operating state of the pumping station 10, which is a facility, and includes water level data from each of the water level gauges 19, 21, and 23, and the control unit 170 described below. There is pump water pumping amount data obtained from the operating state (rotation speed and number of operating pumps) of the pump P controlled by the pump P. Note that the amount of water pumped by the pump may be determined by separately installing a flow meter on the pump P.

The weather data collected by the weather data collection unit 115 is data obtained by inputting weather data distributed on the network at predetermined time intervals. The weather data to be input may be actual measurement data (including short-term forecast data) such as the amount of precipitation at the same time in the area including the treatment area of the pumping station 10, for example, as shown in the weather data 185 shown in FIG. 4(b) below. (The same applies hereinafter), and specifically, analytical rainfall and short-term precipitation forecast data (1 km mesh shape data) provided by the Japan Meteorological Agency are used. As mentioned above, the actual measurement data is meteorological data in mesh units, that is, in area units (1 km mesh shape units) where the area is divided into multiple small areas, meteorological values, longitude / latitude, and time. This is data that links information. The meteorological value is one or more data including precipitation amount, weather, temperature, wind speed, wind direction, and atmospheric pressure. The meteorological value that most influences the prediction of the amount of sewage inflow is the amount of precipitation, but other meteorological values also influence the prediction of the amount of sewage inflow. Therefore, by adding meteorological values other than precipitation to the input values according to the topography and meteorological characteristics of the processing area, it is possible to further improve the accuracy of predicting the amount of sewage inflow.

The storage unit 120 includes a data storage unit 121 , a learned model storage unit 123 , and a geographic information storage unit 125 .

The data storage unit 121 stores and accumulates the facility data and weather data collected by the facility data collection unit 113 and the weather data collection unit 115 over time. The trained model storage unit 123 stores a trained model created by the learning processing unit 150 described below. This learned model is updated as the inflow prediction system 100 is actually operated and new various data are updated. The geographic information storage unit 125 stores geographic information regarding the processing area that is input from the geographic information system GIS distributed over the network.

The learning processing unit 150 is a means for learning a model for predicting the amount of sewage flowing into the pumping station 10 based on the facility data and meteorological data received from the data storage unit 121 to create a learned model. The learning processing unit 150 is a means for creating a model using a machine learning method, and in this example, RNN (Recurrent Neural Network) is used as the machine learning method (more specific (LSTM (Long short-Term Memory), which is an advanced form of RNN, is used for this). RNN is a deep learning method suitable for use in predicting time-series data whose values change over time. Suitable for prediction using facility data and weather data.

In other words, in an RNN, the intermediate layer has a path for transmitting the output from the intermediate layer at a certain time to the intermediate layer at the next time, so that the intermediate layer at time t can transmit the output from the input layer at the same time t. In addition to the input, it also accepts input from the intermediate layer at the previous time t-1. This makes it possible to create a neural network that takes into account time-to-time effects.

In addition, in this embodiment, as one time step of the hyperparameter of the RNN (LSTM), the difference between the peak time of the sewage inflow amount and the peak time of the precipitation amount for the treatment area, and/or the sewage inflow at the sewage treatment plant. A period of fluctuation of the amount (in this example, it was about 24 hours) is used. The amount of sewage inflow at a sewage treatment plant has periodicity specific to each sewage treatment plant (particularly noticeable in sewage treatment plants in urban areas). Since learning and prediction values can be calculated in consideration of the characteristics, calculation costs or the number of learning trials can be reduced, and the accuracy of predicting the amount of sewage inflow can be improved.

The trained model generated by the learning processing unit 150 is stored in the trained model storage unit 123, and is updated at regular intervals.

The prediction unit 160 predicts the amount of sewage inflow at a future point based on the trained model stored in the trained model storage unit 123 and the facility data and weather data stored in the data storage unit 121. The accuracy of the predicted value with respect to the actually measured value of the inflow amount to the pumping station 10 was 0.953 for the correlation coefficient and 216 for the root mean square error (RMSE).

The control unit 170 receives the predicted value of the amount of sewage inflow from the prediction unit 160, and adjusts the opening degree of the gate 25 at the pump station 10, the number of pumps P in operation, and the rotational speed of the pumps P in operation. Adjustment, that is, control the operating state of the pump station 10.

FIG. 3 shows that a trained model of the inflow prediction system 100 is constructed at the pumping station 10, and the trained model is used to predict the inflow of sewage into the pumping station 10. It is a figure showing the outline of one processing procedure which controls the operational state of pumping station 10 based on this. This processing procedure first constructs a trained model that predicts sewage inflow using actual measured values such as facility data, and then, during actual operation of the inflow prediction system 100, uses new actual measured values etc. to learn the model. The operation state of the pump station 10 is controlled while reconstructing the completed model. Note that the order of processing is not limited to this processing procedure, and various changes are possible.

As shown in the figure, in the inflow prediction system 100, a facility data collection unit 113 collects water level data of the inlet culvert 13, settling basin 15, and pump well 17 of the pump station 10, and the water level data of the pump P as facility data PD. Collect pump pumping data (Step 1).

Meanwhile, the weather data collection unit 115 collects weather data 185 (step 2).

Next, the prediction unit 160 uses the collected facility data PD to calculate the past and current inflow amounts of sewage into the pumping station 10 (step 3). That is, the amount of change per unit time in the amount of water held in each water tank, that is, the inlet drain 13, the settling basin 15, and the pump well 17, which changes according to the change in the amount of sewage flowing into the pump station 10, and the unit of the pump P. The sewage inflow amount is calculated by adding up the amount of pumped water per hour and the amount of change per unit time in the amount of storage in the out-of-facility inflow pipe 11 connected to the upstream side of the pumping station 10.

To calculate the amount of change per unit time in the amount of water held in each tank 13, 15, 17, specifically, multiply the cross-sectional area of each tank 13, 15, 17 by the change in water level per unit time. This can be done easily and accurately. In addition, the amount of change per unit time in the amount of storage in the outside inflow pipe 11 is calculated based on the water surface of the inflow culvert 13, which is the water tank to which the outside inflow pipe 11 is connected, and the sewage stored in the outside inflow pipe 11. It can be easily and accurately calculated by assuming that the hydraulic gradient at the water surface is constant and referring to the water level of the inflow conduit 13 and the slope and diameter of the inflow pipe 11 outside the facility. If the amount of sewage inflow is calculated from the amount of fluctuation in the water level in this way, it is possible to calculate the amount of sewage inflow with high accuracy even if a flow meter is not installed to measure the amount of sewage flowing into the pump station 10. I can do it.

Next, the prediction unit 160 uses the previously stored geographic information to calculate weather data for the treatment area where sewage flows into the pumping station 10 (step 4). The procedure for calculating weather data will be explained using FIG. 4.

First, geographic information 181 shown in FIG. 4(a) and weather data 185 shown in FIG. 4(b) are read out. The geographic information 181 stores a treatment area 183 in which a sewer pipe flowing into the pumping station 10 is laid. The processing area 183 is a mesh-shaped processing area 183 whose size (area) matches the mesh size (area) of the weather data 185, but the present invention is not limited to this. Then, as shown in FIG. 4(c), the geographical information 181 and weather data 185 are superimposed, and weather data 185A only for the processing area 183 is extracted as shown in FIG. 4(d). To extract and process this data, for example, we use the clip function in the geographic information system GIS, which allows for integrated analysis of data by visualizing geographic information (map data) 181 on a computer and multiplying it with other information, such as meteorological data 185. use

In this way, among the weather data 185 in mesh units, the weather data 185A at the same time of the mesh portion that overlaps with the area of the treatment area 183 of the pumping station 10, that is, the amount of sewage flowing into the pumping station 10 is determined. The amount of precipitation for each mesh of a certain meteorological data 185A is used as an input value for predicting the amount of sewage inflow. In addition, by using the geographic information system GIS, it is possible to easily and accurately extract only the meteorological data 185A that overlaps the processing area 183 from the meteorological data 185, and from these points, precipitation amount etc. for each mesh and each time can be extracted. weather data can be calculated accurately, and the accurate peak time of precipitation can be calculated by summing up the precipitation amount for each mesh, making it possible to improve the prediction accuracy of sewage inflow.

Using the facility data, weather data, and sewage inflow amount obtained in steps 1 to 4, the learning processing unit 150 is caused to perform learning to construct a trained model that predicts the sewage inflow amount (step 5).

The prediction unit 160 uses the calculated current amount of sewage inflow to the pumping station 10 and the calculated current weather data 185A in the treatment area 183 as explanatory variables in the constructed learned model. The amount of sewage that will flow into the pumping station 10 at a future point (for example, one hour later) is predicted as an objective variable (step 6).

Then, the process moves to step 7, and the current opening degree of the gate 25, the number of pumps P in operation, the rotational speed of each pump P in operation, etc. are controlled based on the predicted value (step 7). The contents of the control include, for example, temporarily increasing the number of operating pumps P or increasing the pump rotation speed based on a predicted value that the amount of sewage flowing into the pump station 10 will increase. The amount of sewage retained in the sewage pipe or pumping station 10 in the treatment area may be decreased in advance by increasing the flow rate of 10. Although not shown, the opening/closing state of another gate installed somewhere along the sewage pipe within the treatment area may be controlled.

Next, when the predetermined time (interval at which the above operation flow is performed) has elapsed (“Y” in step 8), return to step 1, collect and calculate various actual measured values, and re-train the trained model in the same way as above. The system controls gates and pumps based on the estimated sewage inflow volume, and repeats this process to improve the prediction accuracy of the trained model. At this time, in step 5, the trained model may be reconstructed by using not only the actual measured values but also the predicted values, but the trained model may also be reconstructed without necessarily using the predicted values. .

The predetermined time in step 8 is the acquisition interval of the facility data PD and meteorological data 185 used for predicting the amount of sewage inflow, and it is recommended that the predetermined time be set to, for example, an hourly interval, and more preferably a 30-minute interval to improve prediction accuracy. preferable. On the other hand, if the acquisition interval is shortened to, for example, 30 minutes or less, the calculation cost or memory usage for learning in machine learning or outputting predicted values will increase, which may cause problems in actual operation due to a lag in the output of predicted values. There is a possibility that this may occur. On the other hand, if the acquisition interval is increased to 3 hours or more, the prediction accuracy will be lowered, and because the acquisition interval is too long, the meaning of calculating the predicted value will be lost. For these reasons, the acquisition interval is preferably 30 minutes to 3 hours, more preferably 30 minutes to 1 hour.

As explained above, if the inflow prediction system 100 is used, out of the weather data 185 in mesh units, the weather data 185A of the mesh portion that overlaps with the area of the treatment area 183 of the pumping station 10, that is, the inflow into the pumping station 10. Since only the meteorological data 185A that has a high degree of influence on the amount of sewage inflow is used as an input value for predicting the amount of sewage inflow, the accuracy of predicting the amount of sewage inflow can be improved.

The pumping station 10 is composed of an inflow conduit 13, a settling basin 15, and a pump well 17, but the inflow conduit 13 and the settling basin 15 may be omitted depending on the case. Conversely, water tanks other than the above-mentioned water tanks may be installed (the same applies to the pump building 10-2 described below). In either case, the amount of fluctuation per unit time in the amount of water held in each water tank that fluctuates in response to fluctuations in the amount of sewage inflow, the amount of water pumped per unit time, and the amount of water pumped per unit time, and the amount of water that is connected to the upstream side of the pump station 10 are calculated. The amount of sewage inflow is calculated by adding up the amount of change per unit time in the amount of storage in the out-of-facility inflow pipe 11.

Furthermore, in the above processing procedure, a case has been described in which the geographic information system GIS and the meteorological data 185 are input separately to the inflow prediction system 100 and both data are superimposed to calculate meteorological data in the processing area. The system GIS side may take in the meteorological data 185 in advance and process the geographic information so as to overlap it, inputting it into the inflow prediction system 100, and then calculating the meteorological data in the processing area.

FIG. 5 is a schematic configuration diagram showing an example of a sewage treatment plant 200 with a pump building according to another embodiment controlled by the inflow prediction system 100.

The sewage treatment plant 200 with a pump building connects the pump building 10-2, the first settling tank 201, the reaction tank 211, the final settling tank 221, and the disinfection equipment 231 from the upstream side by pipes 17A, 201A, 211A, and 221A, respectively. It is composed of The water tanks may be directly connected without using piping.

Since the pump building 10-2 has the same configuration as the pump station 10 shown in FIG. 1, the same or corresponding parts as those in the pump station 10 shown in FIG.

This pump building 10-2 is different from the pump station 10 described above in that, in addition to the gate 25 and pump P installed in the pump building 10-2, the equipment controlled by the flow rate prediction system 100 is installed in the sewage system with this pump building. This includes the addition of a gate 241 installed at a position where sewage treated at the treatment plant 200 is discharged into a river or the like.

In this inflow prediction system 100, as shown in FIG. weather data 185 (step 1) and simultaneously collect weather data 185 (step 2).

Next, the current inflow of sewage into the pump building 10-2 is calculated using the same method as described above (step 3); Weather data 185A of the treatment area 183 where sewage flows into 10-2 is calculated (step 4).

Next, a trained model for predicting the amount of sewage inflow is constructed using the facility data, weather data, and amount of sewage inflow obtained in steps 1 to 4 (step 5). Then, the constructed trained model is used to predict the amount of sewage inflow (step 6), and based on the predicted value, in addition to the opening/closing status of the gate 25 and the operating status of the pump P, the opening/closing status of the gate 241 is also controlled. (Step 7), the above data acquisition, relearning of the learned model, and prediction are repeated at predetermined time intervals (“Y” in Step 8). For example, based on the predicted value that the amount of sewage flowing into the pump building 10-2 will increase, the content of the control is to increase the number of operating pumps P and the rotation speed in advance, and at the same time change the opening degree of the gates 25, 241. The water level of each water tank (inflow conduit 13, settling tank 15, and pump well 17) in the pump building 10-2, the initial settling tank 201, the reaction tank 211, the final settling tank 221, and the disinfection equipment 231 is lowered in advance. I'm going to have a good time. Thereby, the operation of the sewage treatment plant 200 with a pump building can be controlled in accordance with the amount of sewage inflow predicted in advance.

Note that each water tank constituting the sewage treatment plant with pump building 200 may be omitted depending on the case. Conversely, aquariums other than the above-mentioned aquariums may be installed.

FIG. 6 is a schematic configuration diagram showing an example of a sewage treatment plant 300 without a pump building according to yet another embodiment controlled by the inflow amount prediction system 100.

This sewage treatment plant 300 is configured by connecting, from the upstream side, a first sedimentation tank 301, a reaction tank 311, a final sedimentation tank 321, and a disinfection equipment 331 through pipes 301A, 311A, and 321A, respectively. The water tanks may be directly connected without using piping.

Since this sewage treatment plant 300 does not have a pump building, water level gauges 303, 313, 323, and 333 are installed in each of the initial sedimentation tank 301, reaction tank 311, final sedimentation tank 321, and disinfection equipment 331. The water level gauges 303, 313, 323, and 333 detect changes in the water level of each tank, and each detected value is sent to the inflow prediction system 100 to calculate the amount of sewage flowing into the sewage treatment plant 300. ing.

Further, an outside inflow pipe 351 is connected to the first sedimentation tank 301 of this sewage treatment plant 300. The upstream side of the out-of-facility inflow pipe 351 is connected to a sewage pipe line running around a basin (treatment area) that collects sewage to the sewage treatment plant 300, and sewage from the treatment area is collected. As mentioned above, the collected sewage may be only sewage, only rainwater, or a combination of sewage and rainwater.

A gate 371 is installed at a portion of the initial settling tank 301 connected to the outside inflow pipe 351, and a gate 391 is installed at a position where sewage treated at the sewage treatment plant 300 is discharged into a river or the like. The opening degree of each gate 371, 391 is controlled by the inflow amount prediction system 100.

In this inflow prediction system 100, first, as shown in FIG. 3, water level data of the first sedimentation tank 301, reaction tank 311, final sedimentation tank 321, and disinfection equipment 331 of the sewage treatment plant 300 is collected as facility data PD. (Step 1), and simultaneously collect weather data 185 (Step 2).

Next, the current amount of sewage flowing into the sewage treatment plant 300 is calculated using the collected facility data PD (step 3). Specifically, the unit of water volume held in each water tank, that is, the initial sedimentation tank 301, reaction tank 311, final sedimentation tank 321, and disinfection equipment 331, changes according to changes in the amount of sewage inflow to the sewage treatment plant 300. The amount of change per unit time, the amount of change per unit time of the storage amount in the out-of-facility inflow pipe 351 connected to the upstream side of the sewage treatment plant 300, and in some cases, the piping 301A that connects the respective water tanks, The amount of sewage inflow is calculated by adding up the amount of fluctuation per unit time in the amount of water held in 311A and 321A.

Specifically, the amount of change per unit time in the amount of water held in each tank 301, 311, 321, 331 is calculated by applying This can be done easily and accurately by multiplying by the variation of . Further, the amount of variation per unit time in the storage amount in the outside inflow pipe 351 is calculated with reference to the water level of the water tank 301 to which the outside inflow pipe 351 is connected, and the slope and pipe diameter of the outside inflow pipe 351. This allows easy and accurate determination.

Next, using the same method as described above, meteorological data 185A of the treatment area 183 where sewage flows into the sewage treatment plant 300 is calculated (step 4). Next, a trained model for predicting the amount of sewage inflow is constructed using the facility data, weather data, and amount of sewage inflow obtained in steps 1 to 4 (step 5).

Then, the constructed trained model is used to predict the amount of sewage inflow (step 6), and based on the predicted value, the opening/closing states of the gates 371 and 391 are controlled (step 7), and the above data is collected at predetermined time intervals. acquisition, retraining of the trained model, and prediction are repeated (“Y” in step 8). For example, based on the predicted value that the amount of sewage flowing into the sewage treatment plant 300 will increase, the content of the control is to increase the opening degree of the gates 371 and 391 in advance and adjust the water level of each water tank in the sewage treatment plant 300 in advance. I keep it lowered. Thereby, the operation of this sewage treatment plant 300 can be controlled in accordance with the amount of sewage inflow predicted in advance.

The sewage treatment plant 300 is composed of an initial sedimentation tank 301, a reaction tank 311, a final sedimentation tank 321, and disinfection equipment 331, but any of these tanks may be omitted, or vice versa. In addition, a water tank other than the above-mentioned water tanks may be installed.

FIG. 7 is a schematic plan view when an upstream pump station 500 that collects sewage from an upstream treatment area 501 is connected to one of the sewage pipes in the treatment area 401 of the sewage treatment plant 400.

The upstream pump station 500 is a relay pump station and has the same configuration as the pump station 10 shown in FIG. 1 above. The sewage treatment plant 400 has the same configuration as the sewage treatment plant 200 with a pump building shown in FIG. 5 or the sewage treatment plant 300 shown in FIG. 6. In addition, r1 in the figure represents an image of rainfall for the upstream processing area 501, and r2 represents an image of rainfall for the processing area 401. Since the pump station 500 is located upstream of the sewage treatment plant 400 in this way, the amount of sewage flowing into the sewage treatment plant 400 is affected by the amount of sewage pumped at the pump station 500.

Note that the upstream pump station 500 may instead have the same configuration as the sewage treatment plant 200 with a pump building shown in FIG. 5 or the sewage treatment plant 300 shown in FIG. Alternatively, it may have the same configuration as the pumping station 10 shown in FIG. 1 above. Moreover, the upstream pump station 500 may instead be a pipe facility having an inflow measuring means.

FIG. 8 shows that trained models of the inflow prediction system 100 are constructed for both the sewage treatment plant 400 and the upstream pumping station 500, and these trained models are used to construct the trained models for the sewage treatment plant 400 and the upstream pumping station 500. FIG. 2 is a diagram schematically showing a processing procedure for predicting the amount of sewage flowing into the sewage treatment plant 400 and controlling the operating states of the sewage treatment plant 400 and the upstream pump station 500 based on the prediction. Note that the order of operations is not limited to this processing procedure, and various changes are possible. In this example, the operations of the sewage treatment plant 400 and the upstream pump station 500 are controlled simultaneously by the same inflow prediction system 100, but separate inflow prediction systems 100 are installed for each to control them separately at the same time. It's okay. The inflow rate prediction system 100 used in this example also has the same configuration as the inflow rate prediction system 100 shown in FIG. 2 above.

In FIG. 8, the inflow prediction system 100 first collects facility data PD and weather data 185 at each of the upstream pump station 500 and the sewage treatment plant 400 (steps 1, 1A, 2, 2A).

Next, the amount of sewage inflow and the weather data 185A in the treatment areas 401, 501 are calculated from each of the facility data PD and the weather data 185 (steps 3, 3A, 4, 4A). Next, a trained model for predicting the amount of sewage inflow is constructed using the facility data, weather data, and amount of sewage inflow obtained from steps 1 and 1A to steps 4 and 4A (steps 5 and 5A). In addition, in constructing the learned model used for the sewage treatment plant 400, it is preferable to further use the actual measured value of the amount of water pumped by the pump at the upstream pumping station 500.

Next, in the upstream pumping station 500, the constructed learned model is used to predict the amount of sewage inflow (step 6A), and the predicted value of the sewage inflow is used to pump the pumps of the upstream pumping station 500. A predicted value of the amount of pumped water is calculated (Step 6B). Step 6B of calculating the predicted value of the pumped water amount of the upstream pumping station 500 is referred to as an upstream pumped water amount prediction means.

On the other hand, in the sewage treatment plant 400, the constructed learned model is used to predict the amount of sewage inflow (step 6). In addition to the calculated current weather data 185A in the treatment area 401, the predicted value of the pumped water amount of the upstream pumping station 500 obtained in step 6B (and the current pumped water amount of the upstream pumping station 500) is used for the sewage treatment concerned. The amount of sewage that will flow into the sewage treatment plant 400 at a future point is predicted by inputting it into the learned model for the field 400 (step 6).

With this configuration, even if there is an upstream pump station 500 that collects sewage from the upstream treatment area 501 and flows into the downstream treatment area 401, the predicted value of the pumped water amount of the upstream pump station 500 can be calculated. By inputting the information into the learned model for the sewage treatment plant 400, it becomes possible to predict the amount of sewage inflow into the sewage treatment plant 400 located on the downstream side with high accuracy and ease. That is, by inputting the predicted value of the pumped water amount of the upstream pumping station 500 into the trained model for the sewage treatment plant 400, data from the current pumped water amount of the upstream pumping station 500 to the future pumped water amount can be used for sewage treatment. Since the data can be used as data for predicting the amount of sewage flowing into the field 400, more accurate predictions can be made.

Then, based on each of the above predicted values, the gates and pumps of the upstream pump station 500 and the sewage treatment plant 400 are controlled (steps 7 and 7A), and the learned model is repeatedly retrained and predicted at predetermined time intervals. Continue (“Y” in steps 8 and 8A).

The embodiment shown in FIG. 7 shows a case where an upstream pumping station 500 that collects sewage from an upstream treatment area 501 is connected to a sewage pipe in a treatment area 401 of a sewage treatment plant 400. Connecting the sewage pipe in the treatment area 501 of the pumping station 500 to an upstream pump station that collects sewage from the upstream treatment area further upstream, or connecting a pipe facility having an inflow measurement means, etc. The present invention can be similarly applied to cases where three or more multi-stage processing zones are connected.

When predicting the inflow amount to the sewage treatment plant 400 according to the present invention, if the predicted value of the pumped water amount of the upstream pumping station 500 is not used as an explanatory variable, the predicted value of the inflow amount one hour later will differ from the actual measured value. As for accuracy, the correlation coefficient was 0.914 and the root mean square error (RMSE) was 687. Furthermore, the accuracy of the predicted value of the inflow amount after 10 hours with respect to the actual measured value was 0.719 for the correlation coefficient and 1186 for the root mean square error (RMSE). On the other hand, when the inflow of a pipeline facility having an inflow measuring means and its predicted value are used instead of the upstream pump station 500 as explanatory variables, the accuracy of the predicted value of the inflow after one hour with respect to the actual measured value is , the correlation coefficient was 0.934, and the root mean square error (RMSE) was 615. Furthermore, the accuracy of the predicted value of the inflow amount after 10 hours with respect to the actual measured value was 0.798 for the correlation coefficient and 1051 for the root mean square error (RMSE). In this way, by inputting the predicted value of the pumped water volume of the upstream pump station 500 or the inflow volume of the pipeline facility having an inflow volume measuring means into the trained model for the sewage treatment plant 400, It is possible to improve the accuracy of predicting the amount of inflow.

As described above, in the present invention, a trained model that predicts the amount of sewage inflow is constructed using weather data in mesh units, but it is possible to calculate the degree of influence of the weather data on the amount of sewage inflow in each mesh. can. If the meteorological data includes precipitation and the target of prediction is a separate sewer system, it is also possible to identify locations with a high degree of influence as locations with a high probability of occurrence of unknown water and a high impact. By using the meteorological data of each mesh part that overlaps with the pumping station, the pump building of the sewage treatment plant, the pipeline facility with inflow measurement means, or the treatment area of the sewage treatment plant, more accurate identification can be made. It becomes possible.

Further, in the present invention, the trained model may be rebuilt every fixed period, for example, every six months to five years, preferably every six months to two years. By calculating the degree of influence of precipitation on sewage inflow for each mesh and the change in degree of influence in each period using a trained model built for each fixed period, we If the conduit in the corresponding mesh has progressed to deterioration, it will be possible to take measures such as issuing a warning or prioritizing the repair of the conduit. That is, by setting the period to six months or more, the accuracy of information can be improved, and by setting the period to five years or less, it becomes possible to carry out repairs in a timely manner.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and various modifications can be made within the scope of the claims and the technical idea described in the specification and drawings. It is possible. It should be noted that any structure not directly described in the specification or drawings is within the scope of the technical idea of the present invention as long as it achieves the functions and effects of the present invention. Furthermore, the embodiments described above and shown in the figures can be combined with each other as long as there is no contradiction in purpose, structure, etc. In addition, the above description and the contents of each figure, even if only a part thereof, can be an independent embodiment, and the embodiment of the present invention is an embodiment that combines the above description and each figure. It is not limited to.

10...Pump station, 11...Outside inflow pipe, 13...Inflow drain (water tank), 15...Sand basin (water tank), 17...Pump well (water tank), 19, 21, 23...Water level gauge, 25...Gate, P... Pump, 100...Inflow prediction system, 110...Data collection section, 113...Facility data collection section, 115...Meteorological data collection section, 120...Storage section, 121...Data accumulation section, 123...Learned model storage section, 125... Geographical information storage unit, 150... Learning processing unit, 160... Prediction unit, 170... Control unit, PD... Facility data, GIS... Geographical information system, 181... Geographical information, 183... Processing area, 185, 185A... Meteorological data, 200 ...Sewage treatment plant with pump building, 10-2...Pump building, 201...First sedimentation tank (water tank), 211...Reaction tank (water tank), 221...Final sedimentation tank (water tank), 231...Disinfection equipment (water tank), 241 …Gate, 300…Sewage treatment plant without pump building, 301…First sedimentation tank (water tank), 311…Reaction tank (water tank), 321…Final sedimentation tank (water tank), 331…Disinfection equipment (water tank), 303 , 313, 323, 333...water level gauge, 351...outside inflow pipe, 371,391...gate, 400...sewage treatment plant, 401...treatment area, 500...upstream pump station, 501...upstream treatment area.

Claims (10)

In an inflow rate prediction system that predicts the amount of sewage inflow at a pump station, a pump building in a sewage treatment plant, a pipeline facility having an inflow rate measuring means, or a sewage treatment plant,
a facility data collection unit that collects facility data of the pump station or the pump building in the sewage treatment plant, the pipeline facility having an inflow measurement means, or the sewage treatment plant;
a weather data collection unit that collects weather data;
a data storage unit that stores the facility data and weather data;
A model for predicting the amount of sewage flowing into a pumping station, a pump building in a sewage treatment plant, a pipeline facility having an inflow amount measuring means, or a sewage treatment plant, based on the facility data and meteorological data received from the data storage unit. a learning processing unit that performs learning and creates a trained model;
a trained model storage unit that stores the trained model created by the learning processing unit;
a prediction unit that predicts the amount of sewage inflow at a future point based on the learned model, the facility data, and the weather data;
a control unit that receives a predicted value of the amount of sewage inflow from the prediction unit and controls the operating state of the pump station or the pump building in the sewage treatment plant, or the pipeline facility or the sewage treatment plant that has an inflow amount measuring means;
has
The meteorological data used in the forecasting unit to predict the amount of sewage inflow is meteorological data in mesh units having information on meteorological values, longitude/latitude, and time, and is weather data in mesh units having information on meteorological values, longitude/latitude, and time, and is weather data that is used for predicting the amount of sewage inflow at the pump station or the pump building in the sewage treatment plant or the amount of inflow. An inflow prediction system characterized in that the meteorological data is meteorological data of a mesh portion that overlaps with a treatment area of a pipeline facility or a sewage treatment plant having a measuring means. 2. The inflow amount prediction system according to claim 1, wherein the meteorological value is data including one or more of precipitation amount, weather, temperature, wind speed, wind direction, and atmospheric pressure. 2. The weather data of a mesh portion that overlaps with the processing area is meteorological data of a mesh portion of a region that overlaps with the processing area on geographic information possessed by a geographic information system. inflow prediction system. The inflow amount prediction system according to claim 1, wherein the learning processing unit and the prediction unit perform learning and prediction using a model using machine learning. Using RNN as the machine learning method,
As a time step, which is one of the hyperparameters of the RNN,
The peak time of sewage inflow at the pumping station or sewage treatment plant, the pump building in the sewage treatment plant, or the pipeline facility or sewage treatment plant having a pump building or inflow measurement means, and the pump building or inflow measurement means in the pumping station or sewage treatment plant. The difference between the peak time of precipitation, which is the meteorological value, for the treatment area of the pipeline facility or sewage treatment plant;
and/or the inflow amount according to claim 1, characterized in that a fluctuation cycle of the sewage inflow amount in the pump station, the pump building in the sewage treatment plant, the pipeline facility having an inflow amount measuring means, or the sewage treatment plant is used. Prediction system. The amount of precipitation is calculated based on the amount of precipitation data in mesh units at the same time in the area that overlaps with the area of the pumping station, the pump building in the sewage treatment plant, the pipeline facility having an inflow measurement means, or the treatment area of the sewage treatment plant. It is the total amount of precipitation for each time,
the difference between the peak time of the precipitation and the peak time of the sewage inflow at the pumping station or the pump building in the sewage treatment plant, the pipeline facility having an inflow measurement means, or the sewage treatment plant, and/or the pumping station Alternatively, the time step is determined by calculating a fluctuation period of the sewage inflow in a pump building in a sewage treatment plant, a pipeline facility having an inflow measurement means, or a sewage treatment plant. Prediction system. If the target for predicting the amount of sewage inflow is the pump station or the pump building in the sewage treatment plant,
The amount of sewage inflow is
The amount of fluctuation per unit time in the amount of water held in one or more water tanks of the pump station or pump building, the amount of pumped water per unit time, and the amount of water held in one or more water tanks of the pump station or pump building connected to the upstream side of the pump station or pump building Calculate by adding up the amount of fluctuation per unit time in the amount of sewage stored in the inflow pipes outside the facility,
On the other hand, if the target for predicting the amount of sewage inflow is the sewage treatment plant,
The amount of sewage inflow is
The amount of fluctuation per unit time in the amount of water held in one or more water tanks of the sewage treatment plant, and the amount of sewage stored in the out-of-facility inflow pipe connected to the upstream side of the sewage treatment plant per unit time. 2. The inflow amount prediction system according to claim 1, wherein the inflow amount prediction system is calculated by adding up the amount of variation. The inflow according to claim 7, wherein the amount of change per unit time in the amount of water held in each tank is calculated by multiplying the cross-sectional area of each tank by the change in water level per unit time. Volume prediction system. The amount of variation per unit time in the amount of storage in the inflow pipe outside the facility is:
8. The inflow amount prediction system according to claim 7, wherein the inflow is calculated with reference to the water level of the water tank to which the out-of-facility inflow pipe is connected, and the slope and diameter of the out-of-facility inflow pipe. An upstream pumping station for collecting and inflowing sewage from an upstream treatment area located upstream of the treatment area, or an inflow rate measuring means for an upstream treatment area located upstream of the treatment area, is provided in the treatment area. At least one of the following pipeline facilities is installed,
Using the predicted value of the sewage inflow rate at the upstream pumping station obtained by the inflow rate prediction system, calculate the predicted value of the pumped water rate of the upstream side pumping station, and/or by the inflow rate prediction system. , calculating a predicted value of the amount of sewage inflow in the pipeline facility having an inflow amount measuring means in the upstream treatment area,
The trained model, the facility data and the In addition to the meteorological data, a predicted value of the amount of pumped water at the upstream pumping station and/or a predicted value of the amount of sewage inflow at a pipeline facility having an inflow amount measuring means in the upstream treatment area is used. The inflow amount prediction system according to any one of claims 1 to 9.

Images