Self-adaptive regulation and control method and system for water level of pump station

Abstract

The invention is suitable for the technical field of pump station intelligent control, and provides a pump station water level self-adaptive regulation and control method and system, and the method comprises the following steps: obtaining water level time sequence data, equipment state data and environment information data in real time; predicting a water level value of a future preset time window based on a physical mechanism prediction model according to the water level time sequence data, the equipment state data and the environment information data to obtain a first prediction value; based on a data-driven prediction model, according to the water level time sequence data, the equipment state data and the environment information data, predicting a water level value of a future preset time window to obtain a second prediction value; performing adaptive weighted fusion on the first predicted value and the second predicted value to obtain a comprehensive predicted value; and according to the comprehensive predicted value, carrying out self-adaptive regulation and control on the pump station water level. According to the invention, by constructing a hybrid prediction model based on a physical mechanism and data driving and introducing an adaptive weighted fusion mechanism, accurate short-term prediction and adaptive regulation and control of the future water level are realized.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for pumping stations, and particularly relates to a method and system for adaptive water level control in pumping stations. Background Technology

[0002] Automatic water level control systems for pumping stations are widely used in municipal drainage, reservoir management, and industrial production. Traditional pumping station water level control primarily employs a fixed threshold-based start-stop control method. This involves preset upper and lower water level limits, and when the real-time water level exceeds the set threshold, the corresponding pump start-stop operation is executed. While simple and reliable, this method suffers from significant control lag, failing to anticipate future water level changes. This results in the system only taking action after the water level has exceeded the limit, easily leading to frequent equipment start-stops and drastic water level fluctuations. Especially under sudden conditions such as heavy rain, this passive response mode often cannot cope with rapidly rising water levels in a timely manner, posing safety hazards.

[0003] To improve control performance, some systems employ PID feedback control technology. The PID controller generates a control signal based on the deviation between the setpoint and the measured water level using proportional, integral, and derivative operations. This method offers smoother control compared to simple threshold control, but it still suffers from difficulties in parameter tuning and poor environmental adaptability. Fixed-parameter PID controllers struggle to adapt to complex environmental factors such as variations in rainfall intensity and evaporation rates, and their effectiveness is limited in handling the nonlinearity of pump start-up and shutdown. Furthermore, traditional PID control lacks foresight, and its response to sudden environmental disturbances remains delayed.

[0004] In recent years, with the development of artificial intelligence technology, some research has begun to introduce machine learning methods into water level prediction and control systems. For example, algorithms such as time series analysis, support vector machines, or long short-term memory networks are used for water level prediction, and control strategies are optimized based on the prediction results. However, most of these methods rely on a single data-driven model and lack guidance from physical mechanisms, which may lead to prediction results that violate fundamental physical laws such as the conservation of mass. Summary of the Invention

[0005] The purpose of this invention is to provide a method for adaptive water level control in pumping stations, which aims to solve the above-mentioned technical problems.

[0006] This invention is implemented as follows: a method for adaptive water level control in pumping stations, comprising the following steps:

[0007] Real-time acquisition of water level time-series data, equipment status data, and environmental information data;

[0008] Based on a preset physical mechanism prediction model, the water level value for a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and the first predicted value is obtained.

[0009] Based on a preset data-driven prediction model, the water level value for a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and a second predicted value is obtained.

[0010] The first and second predicted values ​​are adaptively weighted and fused to obtain a comprehensive predicted value.

[0011] Based on comprehensive forecast values, the water level of the pumping station is adaptively adjusted.

[0012] Furthermore, the equipment status data includes the pump's start / stop status, operating power, and characteristic curve; the environmental information data includes the current rainfall intensity, air temperature, water temperature, relative humidity, wind speed, and the predicted rainfall intensity for a future preset time window.

[0013] Furthermore, based on a preset physical mechanism prediction model, and according to water level time series data, equipment status data, and environmental information data, the step of predicting the water level value for a preset time window in the future to obtain the first predicted value specifically includes:

[0014] Based on the law of conservation of mass, a physical mechanism prediction model is constructed.

[0015] The inflow rate is determined based on the current rainfall intensity and the predicted rainfall intensity for the next preset time window;

[0016] The outflow rate is determined based on water level time series data, equipment status data, and environmental information data;

[0017] Based on the physical mechanism prediction model, the water level value for a future preset time window is predicted according to the current water level, inflow rate, and outflow rate, thus obtaining the first predicted value.

[0018] Furthermore, the step of determining the outflow rate based on water level time-series data, equipment status data, and environmental information data specifically includes:

[0019] Calculate the pump discharge flow rate by analyzing the pump's start / stop status, operating power, and characteristic curves.

[0020] Calculate the evaporation loss based on the current air temperature, water temperature, relative humidity, and wind speed;

[0021] Calculate leakage loss based on the current water level;

[0022] The outflow rate is determined based on the pump outflow rate, evaporation loss, and leakage loss.

[0023] Furthermore, based on a preset data-driven prediction model, and according to water level time-series data, equipment status data, and environmental information data, the step of predicting the water level value for a preset time window in the future to obtain a second predicted value specifically includes:

[0024] Based on deep learning networks, build and train data-driven prediction models;

[0025] Based on the real-time water level time series data, determine the current water level value, the first derivative and the second derivative of the water level time series data;

[0026] The current water level, the first and second derivatives of the water level time series data, equipment status data, and environmental information data are used as input features and fed into the data-driven prediction model to predict the water level value in the future preset time window and output a second predicted value.

[0027] Furthermore, the step of adaptively weighting and fusing the first and second predicted values ​​to obtain a comprehensive predicted value specifically includes:

[0028] Based on the recent prediction errors of the physical mechanism prediction model and the data-driven prediction model, and combined with the current rainfall intensity and the predicted rainfall intensity for the future preset time window, the fusion weights are dynamically adjusted.

[0029] Based on the dynamically adjusted fusion weights, the first predicted value and the second predicted value are weighted and fused to obtain the comprehensive predicted value.

[0030] Furthermore, the steps for adaptively regulating the pump station water level based on comprehensive forecast values ​​specifically include:

[0031] The water level deviation is determined based on the comprehensive forecast value and the preset target water level value;

[0032] Based on the PID control algorithm, the water level of the pumping station is adaptively adjusted according to the water level deviation.

[0033] Another object of the present invention is to provide a pumping station water level adaptive control system for implementing the above-mentioned pumping station water level adaptive control method, comprising:

[0034] The real-time data acquisition module is used to acquire water level time-series data, equipment status data, and environmental information data in real time.

[0035] The first prediction module is used to predict the water level value in a future preset time window based on a preset physical mechanism prediction model, according to water level time series data, equipment status data and environmental information data, and obtain the first prediction value.

[0036] The second prediction module is used to predict the water level value in a future preset time window based on a preset data-driven prediction model, according to water level time series data, equipment status data and environmental information data, and obtain the second prediction value.

[0037] The weighted fusion module is used to adaptively weight and fuse the first predicted value and the second predicted value to obtain a comprehensive predicted value;

[0038] The water level control module is used to adaptively control the water level of the pumping station based on comprehensive forecast values.

[0039] Furthermore, the first prediction module specifically includes:

[0040] The first model building unit is used to build a physical mechanism prediction model based on the law of conservation of mass.

[0041] The inflow determination unit is used to determine the inflow based on the current rainfall intensity and the predicted rainfall intensity for a future preset time window;

[0042] The outflow rate determination unit is used to determine the outflow rate based on water level time series data, equipment status data, and environmental information data.

[0043] The first prediction value output unit is used to predict the water level value of a future preset time window based on the physical mechanism prediction model, according to the current water level value, inflow rate and outflow rate, to obtain the first prediction value.

[0044] Furthermore, the second prediction module includes:

[0045] The second model building unit is used to build and train data-driven prediction models based on deep learning networks.

[0046] The feature advance unit is used to determine the current water level value, the first derivative and the second derivative of the water level time series data based on the real-time acquired water level time series data;

[0047] The second prediction value output unit is used to input the current water level value, the first and second derivatives of the water level time series data, equipment status data, and environmental information data as input features into the data-driven prediction model to predict the water level value in the future preset time window and output the second prediction value.

[0048] This invention provides a method for adaptive water level control in pumping stations. By constructing a hybrid prediction model based on physical mechanisms and data-driven approaches, and introducing an adaptive weighted fusion mechanism, it achieves accurate short-term prediction and adaptive control of future water levels. This effectively overcomes the lag, rigidity, and isolation of traditional water level control methods, and provides reliable technical support for smart water systems. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating a method for adaptive water level control in a pumping station, as provided in an embodiment of the present invention.

[0050] Figure 2 This is a flowchart illustrating step S200 in a pump station water level adaptive control method provided in an embodiment of the present invention.

[0051] Figure 3 This is a flowchart illustrating step S230 of a pump station water level adaptive control method provided in an embodiment of the present invention.

[0052] Figure 4 This is a flowchart illustrating step S300 in a pump station water level adaptive control method provided in an embodiment of the present invention.

[0053] Figure 5 This is a flowchart illustrating step S400 in a pump station water level adaptive control method provided in an embodiment of the present invention.

[0054] Figure 6 This is a flowchart illustrating step S500 in a pump station water level adaptive control method provided in an embodiment of the present invention.

[0055] Figure 7 This is a schematic diagram of a pump station water level adaptive control system provided in an embodiment of the present invention.

[0056] Figure 8 This is a schematic diagram of the structure of the first prediction module provided in an embodiment of the present invention.

[0057] Figure 9 This is a schematic diagram of the structure of the second prediction module provided in an embodiment of the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0059] like Figure 1 As shown, in one embodiment of the present invention, a method for adaptive water level control in a pumping station is provided, comprising the following steps:

[0060] S100: Real-time acquisition of water level time-series data, equipment status data, and environmental information data;

[0061] S200: Based on a preset physical mechanism prediction model, the water level value for a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and the first predicted value is obtained.

[0062] S300: Based on a preset data-driven prediction model, the water level value of a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and a second predicted value is obtained.

[0063] S400. Adaptively weighted and fused the first predicted value and the second predicted value to obtain a comprehensive predicted value;

[0064] S500: Based on comprehensive forecast values, the water level of the pumping station is adaptively controlled.

[0065] Among them, water level time series data refers to the current water level value and the water level value of a certain time window (such as 10-60min) arranged in the order of timestamps; equipment status data includes pump start-up and shutdown status, operating power and characteristic curve (flow-power curve) and other data; environmental information data includes current rainfall intensity, air temperature, water temperature, relative humidity, wind speed and the predicted rainfall intensity value of a future preset time window (such as 1-10min) (which can be obtained from weather forecasts from meteorological stations) and other data.

[0066] like Figure 2 As shown, in a preferred embodiment of the present invention, the step of predicting the water level value of a future preset time window based on a preset physical mechanism prediction model, according to water level time series data, equipment status data, and environmental information data, and obtaining a first predicted value, namely step S200, specifically includes:

[0067] S210. Based on the law of conservation of mass, construct a physical mechanism prediction model; the specific physical mechanism prediction model is as follows:

[0068] ;

[0069] Where H1 is the first predicted value, typically in meters (m); H0 is the current water level, typically in millimeters (mm); A s The cross-sectional area of ​​the pump station's water storage area (generally taken as the water surface area A of the pump station's water storage area). w and base area A d (average value), usually in mm. 2 ; This is a preset time window, typically ranging from 1 to 10 minutes; Q in The flow rate flowing into the pumping station's water storage area is called the inflow flow rate, typically expressed in mm. 3 / min; Q out The outflow rate is the flow rate from the water storage area of ​​the pumping station, typically expressed in mm. 3 / min.

[0070] S220. Determine the inflow rate based on the current rainfall intensity and the predicted rainfall intensity for the next preset time window;

[0071] Specifically, inflow mainly includes the flow from rainfall, and depending on the actual situation, it may also include other inflows such as groundwater, sewage, and rivers; the formula for calculating inflow is as follows:

[0072] ;

[0073] ;

[0074] Among them, Q r The flow rate resulting from rainfall, typically measured in mm³ / min; O in Other inflows, typically expressed in mm. 3 / min; R c The current rainfall intensity is typically expressed in mm / min; R p The predicted rainfall intensity for a future time window, typically expressed in mm / min; A c The unit is the catchment area, usually in mm. 2 ; This is the runoff coefficient, which is dimensionless and can be determined based on the actual land surface type, generally taking a value of 0-1.

[0075] S230. Determine the outflow rate based on water level time series data, equipment status data, and environmental information data;

[0076] S240. Based on the physical mechanism prediction model, predict the water level value in the future preset time window according to the current water level value, inflow rate and outflow rate, and obtain the first predicted value.

[0077] like Figure 3 As shown, in a preferred embodiment of the present invention, the step of determining the outflow rate based on water level time-series data, equipment status data, and environmental information data, i.e., step S230, specifically includes:

[0078] S231. Calculate the pump discharge flow rate by considering the pump's start / stop status, operating power, and characteristic curve; the formula for calculating the pump discharge flow rate is as follows:

[0079] ;

[0080] Among them, Q p This refers to the pump discharge flow rate, typically expressed in mm. 3 / min; Let P be the start / stop state variable of the i-th pump, which is 1 when the pump starts and 0 when the pump stops; i Let be the current operating power of the i-th pump; The function expression for the power-flow rate characteristic curve of the i-th pump is provided by the pump manufacturer; N is the total number of pumps in the pumping station.

[0081] S232. Calculate the evaporation loss using the current air temperature, water temperature, relative humidity, and wind speed; the formula for calculating the evaporation loss is as follows:

[0082] ;

[0083] Among them, Q e Evaporation loss, typically expressed in mm. 3 / min; v is the wind speed, usually in mm / min; Temperature T a and water temperature T w The difference, usually in K; E s (T w E represents the saturated vapor pressure corresponding to the water temperature, which is related to the water temperature and can be calculated using the Antoine equation. The unit is generally kPa. a The pressure is the actual water vapor pressure, which is related to relative humidity, and the unit is generally kPa; k1, k2, and k3 are all empirical coefficients, which can be determined by experiments and the least squares method. The unit of k1 is generally mm / (min·kPa), and the unit of k2 is generally kPa. -1 The unit of k3 is generally mm / (min·kPa·K); A w The water surface area of ​​the pump station's water storage area, typically expressed in mm. 2 .

[0084] S233. Calculate the leakage loss based on the current water level; the formula for calculating leakage loss is as follows:

[0085] ;

[0086] Among them, Q l This refers to leakage loss, typically expressed in mm. 3 / min; The permeability coefficient is related to the properties of the permeable medium and fluid, and can be determined based on practical experience; its unit is generally mm / min. L is the seepage path length, i.e., the thickness of the permeable layer (such as soil), and its unit is generally mm. A d This refers to the bottom surface area of ​​the pump station's water storage area, typically expressed in mm. 2 .

[0087] S234. Determine the outflow rate based on the pump output flow rate, evaporation loss, and leakage loss; the formula for calculating the outflow rate is as follows:

[0088] ;

[0089] Among them, Qout The outflow rate mainly includes the pump outflow rate Q. p Evaporation loss Q e and leakage loss Q l Depending on the actual situation, other lost traffic, such as natural overflow, may also be included. out .

[0090] like Figure 4 As shown, in a preferred embodiment of the present invention, the step of predicting the water level value for a future preset time window based on a preset data-driven prediction model, according to water level time series data, equipment status data, and environmental information data, and obtaining a second predicted value, namely step S300, specifically includes:

[0091] S310. Based on deep learning networks, construct and train data-driven prediction models;

[0092] Specifically, historical water level time-series data, equipment status data, and environmental information data are first acquired and timestamped, and data from different sampling frequencies are processed. Next, preprocessing such as data cleaning and anomaly removal is performed. Feature extraction is then performed on the preprocessed data. Specifically, features such as water level values ​​at various times, the first and second derivatives of the water level time-series data are extracted from the water level time-series data; features such as pump start / stop status, operating power, and characteristic curves are extracted from the equipment status data; and features such as real-time rainfall intensity, air temperature, water temperature, relative humidity, wind speed, and predicted rainfall intensity for a preset future time window are extracted from the environmental information data. Neural networks such as TCN, LSTM, or Transformer are selected as the model architecture, using the above features as input features and the water position at a certain future time as the output. MSE is used as the loss function for model training, and Bayesian optimization is used for hyperparameter optimization to obtain the optimal data-driven prediction model.

[0093] S320. Based on the real-time water level time series data, determine the current water level value, the first derivative and the second derivative of the water level time series data;

[0094] S330. The current water level value, the first and second derivatives of the water level time series data, equipment status data (including pump start / stop status, operating power, and characteristic curves), and environmental information data (including current rainfall intensity, air temperature, water temperature, relative humidity, wind speed, and the predicted rainfall intensity for a future preset time window) are input features and fed into the data-driven prediction model to predict the water level value for the future preset time window and output a second predicted value.

[0095] like Figure 5 As shown, in a preferred embodiment of the present invention, the step of adaptively weighting and fusing the first predicted value and the second predicted value to obtain a comprehensive predicted value, i.e., step S400, specifically includes:

[0096] S410. Based on the recent prediction errors of the physical mechanism prediction model and the data-driven prediction model, and combined with the current rainfall intensity and the predicted rainfall intensity value for the future preset time window, dynamically adjust the fusion weights.

[0097] S420. Based on the dynamically adjusted fusion weights, the first predicted value and the second predicted value are weighted and fused to obtain the comprehensive predicted value.

[0098] Specifically, the weighted fusion formula for the comprehensive predicted values ​​is as follows:

[0099] ;

[0100] in, The fusion weights are: 1 - w = 1 / w, 1 / w = 1 / w, and 1 / w = 1 / w. The fusion weights of the physics-mechanism prediction model and the data-driven prediction model are dynamically adjusted based on the recent prediction errors (absolute values) of both models, combined with the current rainfall intensity and the predicted rainfall intensity for a future preset time window. For example, during periods of drastic rainfall intensity changes (such as when a sudden rainstorm is expected), the fusion weights of the data-driven prediction model increase; during periods of stability, the fusion weights of the physics-mechanism prediction model increase. The adjustment formula for the fusion weights is as follows:

[0101] ;

[0102] in, This represents the average absolute value of the recent prediction errors of the physical mechanism prediction model; The average absolute value of the recent prediction error of the data-driven prediction model; This is a correction factor, related to the rate of change of rainfall intensity. Generally speaking, when the rate of change of rainfall intensity is zero, The value can be 0.5; when the rate of change of rainfall intensity is greater than or equal to the preset threshold (which can be determined according to the actual situation), The value can be 0.9; when the rate of change of rainfall intensity is less than the preset threshold and greater than zero, Possible value: 0.7

[0103] like Figure 6 As shown, in a preferred embodiment of the present invention, the step of adaptively regulating the water level of the pumping station based on the comprehensive predicted value, i.e., step S500, specifically includes:

[0104] S510. Determine the water level deviation based on the comprehensive forecast value and the preset target water level value; the formula for calculating the water level deviation is as follows:

[0105] ;

[0106] Where e is the water level deviation; H s The target water level can be determined based on the actual conditions of the pumping station, or it can be dynamically adjusted according to demand and environmental information.

[0107] S520, based on PID control algorithm, adaptively regulates the water level of the pumping station according to the water level deviation;

[0108] Specifically, the proportional, integral, and derivative coefficients of the PID control algorithm, combined with the water level deviation, can be used to calculate the pump speed or power adjustment value, thereby controlling the water level of the pumping station. For example, when the comprehensive predicted value is greater than the target water level, the pump speed or power adjustment value is positive, and the PID control algorithm can automatically increase the pump speed or power to increase the outflow and lower the water level. Conversely, when the comprehensive predicted value is less than the target water level, the pump speed or power adjustment value is negative, and the PID control algorithm can automatically reduce the pump speed or power to decrease the outflow and raise the water level. Furthermore, when the comprehensive predicted value equals the target water level, the pump speed or power adjustment value is zero, and the current pump state can be maintained.

[0109] In summary, the pump station water level adaptive control method provided by the embodiments of the present invention, by constructing a hybrid prediction model based on physical mechanisms and data-driven approaches, and introducing an adaptive weighted fusion mechanism, achieves accurate short-term prediction and adaptive control of future water levels. This effectively overcomes the lag, rigidity, and isolation of traditional water level control methods, and provides reliable technical support for smart water systems.

[0110] like Figure 7 As shown, in another embodiment of the present invention, a pumping station water level adaptive control system is also provided to implement the above-mentioned pumping station water level adaptive control method, comprising:

[0111] The real-time data acquisition module 10 is used to acquire water level time-series data, equipment status data, and environmental information data in real time.

[0112] The first prediction module 20 is used to predict the water level value in a future preset time window based on a preset physical mechanism prediction model, according to water level time series data, equipment status data and environmental information data, to obtain the first prediction value.

[0113] The second prediction module 30 is used to predict the water level value of a future preset time window based on a preset data-driven prediction model, according to water level time series data, equipment status data and environmental information data, and to obtain a second prediction value.

[0114] The weighted fusion module 40 is used to adaptively weight and fuse the first predicted value and the second predicted value to obtain a comprehensive predicted value.

[0115] The water level control module 50 is used to adaptively control the water level of the pumping station based on comprehensive prediction values.

[0116] like Figure 8 As shown, in a preferred embodiment of the present invention, the first prediction module 20 specifically includes:

[0117] The first model building unit 21 is used to build a physical mechanism prediction model based on the law of conservation of mass.

[0118] The inflow determination unit 22 is used to determine the inflow based on the current rainfall intensity and the predicted rainfall intensity for a future preset time window;

[0119] The outflow rate determination unit 23 is used to determine the outflow rate based on water level time series data, equipment status data, and environmental information data;

[0120] The first prediction value output unit 24 is used to predict the water level value of a future preset time window based on the physical mechanism prediction model, according to the current water level value, inflow rate and outflow rate, to obtain the first prediction value.

[0121] like Figure 9 As shown, in a preferred embodiment of the present invention, the second prediction module 30 includes:

[0122] The second model building unit 31 is used to build and train a data-driven prediction model based on a deep learning network.

[0123] The feature advance unit 32 is used to determine the current water level value, the first derivative and the second derivative of the water level time series data based on the real-time acquired water level time series data;

[0124] The second prediction value output unit 33 is used to input the current water level value, the first and second derivatives of the water level time series data, the equipment status data, and the environmental information data as input features into the data-driven prediction model to predict the water level value in the future preset time window and output the second prediction value.

[0125] It should be noted that each of the above modules or units can be implemented as a computer program, which can run on a computer device. The computer device's memory can store the computer program that makes up each module, enabling the processor to execute each step of the above method.

[0126] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0127] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory.

[0128] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for adaptive water level control in a pumping station, characterized in that, Includes the following steps: Real-time acquisition of water level time-series data, equipment status data, and environmental information data; Based on a preset physical mechanism prediction model, the water level value for a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and the first predicted value is obtained. Based on a preset data-driven prediction model, the water level value for a preset time window is predicted according to water level time series data, equipment status data and environmental information data, and a second predicted value is obtained. The first and second predicted values ​​are adaptively weighted and fused to obtain a comprehensive predicted value. Based on comprehensive forecast values, the water level of the pumping station is adaptively adjusted.

2. The pump station water level adaptive control method according to claim 1, characterized in that, The equipment status data includes the pump's start / stop status, operating power, and characteristic curves; the environmental information data includes the current rainfall intensity, air temperature, water temperature, relative humidity, wind speed, and the predicted rainfall intensity for a future preset time window.

3. The pump station water level adaptive control method according to claim 2, characterized in that, Based on a pre-defined physical mechanism prediction model, and using water level time-series data, equipment status data, and environmental information data, the steps for predicting the water level value within a pre-defined time window to obtain the first predicted value specifically include: Based on the law of conservation of mass, a physical mechanism prediction model is constructed. The inflow rate is determined based on the current rainfall intensity and the predicted rainfall intensity for the next preset time window; The outflow rate is determined based on water level time series data, equipment status data, and environmental information data; Based on the physical mechanism prediction model, the water level value for a future preset time window is predicted according to the current water level, inflow rate, and outflow rate, thus obtaining the first predicted value.

4. The pump station water level adaptive control method according to claim 3, characterized in that, The step of determining the outflow rate based on water level time-series data, equipment status data, and environmental information data specifically includes: Calculate the pump discharge flow rate by analyzing the pump's start / stop status, operating power, and characteristic curves. Calculate the evaporation loss based on the current air temperature, water temperature, relative humidity, and wind speed; Calculate leakage loss based on the current water level; The outflow rate is determined based on the pump outflow rate, evaporation loss, and leakage loss.

5. The pump station water level adaptive control method according to claim 2, characterized in that, Based on a preset data-driven prediction model, the steps of predicting the water level value for a preset time window and obtaining a second predicted value according to water level time series data, equipment status data, and environmental information data specifically include: Based on deep learning networks, build and train data-driven prediction models; Based on the real-time water level time series data, determine the current water level value, the first derivative and the second derivative of the water level time series data; The current water level, the first and second derivatives of the water level time series data, equipment status data, and environmental information data are used as input features and fed into the data-driven prediction model to predict the water level value in the future preset time window and output a second predicted value.

6. The pump station water level adaptive control method according to claim 2, characterized in that, The steps for adaptively weighting and fusing the first and second predicted values ​​to obtain a comprehensive predicted value include: Based on the recent prediction errors of the physical mechanism prediction model and the data-driven prediction model, and combined with the current rainfall intensity and the predicted rainfall intensity for the future preset time window, the fusion weights are dynamically adjusted. Based on the dynamically adjusted fusion weights, the first predicted value and the second predicted value are weighted and fused to obtain the comprehensive predicted value.

7. The pump station water level adaptive control method according to claim 1, characterized in that, The steps for adaptively regulating the water level of the pumping station based on comprehensive forecast values ​​specifically include: The water level deviation is determined based on the comprehensive forecast value and the preset target water level value; Based on the PID control algorithm, the water level of the pumping station is adaptively adjusted according to the water level deviation.

8. A pumping station water level adaptive control system, used to implement the pumping station water level adaptive control method according to any one of claims 1-7, characterized in that, include: The real-time data acquisition module is used to acquire water level time-series data, equipment status data, and environmental information data in real time. The first prediction module is used to predict the water level value in a future preset time window based on a preset physical mechanism prediction model, according to water level time series data, equipment status data and environmental information data, and obtain the first prediction value. The second prediction module is used to predict the water level value in a future preset time window based on a preset data-driven prediction model, according to water level time series data, equipment status data and environmental information data, and obtain the second prediction value. The weighted fusion module is used to adaptively weight and fuse the first predicted value and the second predicted value to obtain a comprehensive predicted value; The water level control module is used to adaptively control the water level of the pumping station based on comprehensive forecast values.

9. The pump station water level adaptive control system according to claim 8, characterized in that, The first prediction module specifically includes: The first model building unit is used to build a physical mechanism prediction model based on the law of conservation of mass. The inflow determination unit is used to determine the inflow based on the current rainfall intensity and the predicted rainfall intensity for a future preset time window; The outflow rate determination unit is used to determine the outflow rate based on water level time series data, equipment status data, and environmental information data. The first prediction value output unit is used to predict the water level value of a future preset time window based on the physical mechanism prediction model, according to the current water level value, inflow rate and outflow rate, to obtain the first prediction value.

10. The pump station water level adaptive control system according to claim 8, characterized in that, The second prediction module includes: The second model building unit is used to build and train data-driven prediction models based on deep learning networks. The feature advance unit is used to determine the current water level value, the first derivative and the second derivative of the water level time series data based on the real-time acquired water level time series data; The second prediction value output unit is used to input the current water level value, the first and second derivatives of the water level time series data, equipment status data, and environmental information data as input features into the data-driven prediction model to predict the water level value in the future preset time window and output the second prediction value.

Images