Data-driven scheduling optimization method and terminal equipment for cascade pumping stations

Abstract

The present invention belongs to the technical field of cascade pump station scheduling optimization, and specifically discloses a data-driven cascade pump station scheduling optimization method and terminal equipment. The method includes: presetting the target water demand of the final pump station and the target water level of each pump station outlet tank at a preset time; determining the neighborhood output tracking error of each pump station outlet tank based on the target water demand and the target water level; determining the estimated value of the time-varying parameter between the input error and the output error of the outlet tank of each pump station; and determining the water level of each pump station inlet tank at a given time based on the estimated value of the time-varying parameter, the neighborhood output tracking error, and the water level of each pump station inlet tank at a given time. The present invention dynamically adjusts the local control strategy based on the time-varying parameter, can realize adaptive control of a single pump station, and a single pump station can follow the target water demand of the final pump station to achieve the desired control. The present invention can realize consistent control of cascade pump stations without relying on a model, reduce the complexity of scheduling optimization, and have high robustness.

Description

Technical Field

[0001] The present invention belongs to the technical field of cascade pump station scheduling optimization, and specifically discloses a data-driven cascade pump station scheduling optimization method and terminal equipment. Background Art

[0002] Water resources play an irreplaceable role in industry, agriculture, and social life. However, water resources are often distributed extremely unevenly. To address regional water shortages, cascade pumping station projects are often constructed in areas with relatively scarce water resources, creating an operational system that meets regional water supply requirements while optimizing energy consumption. Traditional cascade pumping station scheduling methods largely rely on engineering pre-sets and manual scheduling, which are significantly influenced by subjective factors. This results in insufficient system efficiency and leaves room for improvement. Therefore, how to achieve both efficient and precise scheduling of cascade pumping stations to achieve energy conservation, consumption reduction, and improved economic benefits has become a pressing issue.

[0003] To solve the above problems, current cascade pump station scheduling often adopts a model-based control strategy. However, faced with complex hydraulic conditions and real-time dynamic changing factors, the model-based cascade pump station scheduling method faces modeling difficulties, resulting in the overall operating efficiency being difficult to achieve the optimal state, and it is difficult to achieve global coordinated control technical problems. Summary of the Invention

[0004] The purpose of the present invention is to provide a data-driven cascade pump station scheduling optimization method and terminal equipment to solve the technical problems faced by existing model-based cascade pump station scheduling methods, which make it difficult to achieve the optimal overall operating efficiency and difficult to achieve global coordinated control.

[0005] A first aspect of the present invention provides a data-driven cascade pump station scheduling optimization method, comprising:

[0006] Step 1. Preset The target water demand of the last pumping station at the moment and the target water level of the outlet pool of each pumping station in the cascade pumping station.

[0007] Step 2: Determine the neighborhood output tracking error of each pump station outlet pool according to the target water demand and the target water level.

[0008] Step 3: Identify each pump station The input error of the water tank at the moment is The estimated value of the time-varying parameter between the output error of the water pool at each moment.

[0009] Step 4: Output tracking error according to the estimated value of the time-varying parameter, the neighborhood and The water level of each pump station inlet pool at the moment is determined The water level of the water inlet pool of each pumping station at all times.

[0010] Preferably, the step 2 is specifically as follows:

[0011] Step 2.1: Construct a communication topology graph of the cascade pumping station and determine the adjacency matrix of the communication topology graph.

[0012] Step 2.2: Determine the error coefficient of each pumping station based on the adjacency matrix and the leadership relationship between the final pumping station and each of the remaining pumping stations.

[0013] Step 2.3: Determine the neighborhood output tracking error of each pump station outlet pool based on the error coefficient, the target water demand, and the target water level.

[0014] Preferably, the error coefficient includes Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations , , is the total number of pumping stations in the cascade pumping station; then step 2.2 is specifically as follows:

[0015] When When a pumping station directly accepts the leadership of the final pumping station, , .

[0016] otherwise , , is the element of the adjacency matrix.

[0017] Preferably, step 2.3 is specifically:

[0018] Determine the target water demand and The first difference in target water levels for each pumping station.

[0019] Determine the The target water level of each pump station is The second difference in target water levels for each pumping station.

[0020] Determine the first difference, the second difference and the error coefficient The tracking error of the neighborhood output of the water outlet pool of each pumping station.

[0021] Preferably, the first difference, the second difference and the error coefficient are used to determine the first The tracking error of the neighborhood output of the water outlet pool of each pump station is:

[0022] Determine the first difference and The product of is recorded as the first value.

[0023] Determine the second difference and The product of the second value is recorded, and the second value of the second value is determined by the product of the second value of the second value. The sum of the second values ​​of the pumping stations is recorded as the third value.

[0024] Determine the first value based on the sum of the first value and the third value The tracking error of the neighborhood output of the water outlet pool of each pumping station.

[0025] Preferably, step 4 is specifically:

[0026] An incremental value is determined according to the estimated value of the time-varying parameter and the neighborhood output tracking error.

[0027] According to the increment value and The water level of each pump station inlet pool is determined at the moment The water level of the water inlet pool of each pumping station at all times.

[0028] Preferably, the incremental value is determined according to the estimated value of the time-varying parameter and the neighborhood output tracking error, specifically:

[0029] Get the step size factor and regularization parameter when updating the water level of the inlet pool of each pumping station.

[0030] An incremental value is determined according to the step size factor, the regularization parameter, the estimated value of the time-varying parameter, and the neighborhood output tracking error.

[0031] Preferably, the step size factor is based on Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations Sure.

[0032] Preferably, the regularization parameter is determined according to the average value of the time-varying parameter on the time axis.

[0033] The second aspect of the present invention provides a terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned data-driven cascade pump station scheduling optimization method when executing the computer program.

[0034] Compared with the prior art, the data-driven cascade pump station scheduling optimization method and terminal device of the present invention have the following beneficial effects:

[0035] This invention dynamically adjusts local control strategies based on time-varying parameters determined by real-time sampled input and output data. This enables adaptive control of individual pumping stations, ensuring that each station follows the target water demand of the final pumping station to achieve desired control. This invention achieves consistent control of cascade pumping stations without relying on a model, reducing scheduling optimization complexity and improving robustness.

[0036] The present invention optimizes the internal relationships of cascade pumping stations by improving tracking errors, solves the problem that some pumping stations cannot directly access the last-stage pumping station, makes the scheduling of cascade pumping stations more accurate, and improves the operating efficiency and economic benefits of cascade pumping stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of a data-driven cascade pump station scheduling optimization method according to an embodiment of the present invention.

[0038] Figure 2 This is a communication topology diagram of a cascade pump station according to a specific embodiment of the present invention.

[0039] Figure 3 This is a diagram showing the water level coordinated control results determined using the traditional model-free adaptive control algorithm.

[0040] Figure 4 This is a diagram of the water level coordination control results determined by the data-driven cascade pump station scheduling optimization method of the present invention. DETAILED DESCRIPTION

[0041] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0042] The first aspect of the embodiment of the present invention provides a data-driven cascade pump station scheduling optimization method, such as Figure 1 Shown, including:

[0043] Step 1. Preset Target water demand of the final pumping station at any given moment and the target water level of each pump station outlet pool in the cascade pump station , , is the total number of pumping stations in the cascade pumping station.

[0044] The embodiment of the present invention will As the expected output and reference state of the overall operation of the cascade pumping station, it serves as the leader trajectory in the subsequent communication topology graph.

[0045] Step 2: According to target water demand and target water level Determine the neighborhood output tracking error for each pumping station outlet pool , specifically:

[0046] Step 2.1: Construct a communication topology graph of the cascade pumping station and determine the adjacency matrix of the communication topology graph.

[0047] In the embodiment of the present invention, each pumping station of the cascade pumping station is regarded as an intelligent agent, and a directed communication topology diagram between the intelligent agents is determined according to the mutual relationship between the upstream and downstream pumping stations.

[0048] For example, a directed communication topology graph is defined .

[0049] in, It is the set of all agent vertices in the communication topology graph; represents the communication between agents, Description of the Agent Able to receive intelligent information; Communication topology diagram The adjacency matrix of hour, ,otherwise .

[0050] The embodiment of the present invention further defines the leader adjacency matrix ,in, Representing an agent Can directly receive information from the leader, Representing an agent Cannot receive information from the leader directly.

[0051] The definition of the embodiment of the present invention is intended to mathematically model the topological relationship between various intelligent agents and provide a mathematical basis for stability analysis and control scheme design.

[0052] Step 2.2, according to the adjacency matrix 、 The error coefficient of each pumping station is determined by the leadership relationship between the final pumping station and each of the remaining pumping stations. and .

[0053] The error coefficients in the embodiment of the present invention include Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations , , is the total number of pumping stations in the cascade pumping station; then step 2.2 is specifically as follows:

[0054] When When a pumping station directly accepts the leadership of the final pumping station, According to the adjacency matrix Sure, , ;otherwise , , is the adjacency matrix elements.

[0055] Step 2.3, according to the error coefficient 、 Target water demand and target water level Determine the neighborhood output tracking error for each pumping station outlet pool , specifically: Determine target water demand With the Target water level of each pumping station The first difference of Target water level of each pumping station With the Target water level of each pumping station The second difference; according to the first difference, the second difference and the error coefficient 、 Determine the Neighborhood output tracking error of the pump station outlet pool .

[0056] The above is based on the first difference, the second difference and the error coefficient 、 Determine the Neighborhood output tracking error of the pump station outlet pool , specifically: determine the first difference and The product of is recorded as the first value; determine the second difference and The product of the second value is recorded, and the second value of the second value is determined by the product of the second value of the second value. The sum of the second values ​​of the pumping stations is recorded as the third value; the sum of the first and third values ​​is used to determine the Neighborhood output tracking error of the outlet pool of the secondary pumping station .

[0057] For example, the following formula (1) is used to determine Neighborhood output tracking error of the pump station outlet pool :

[0058] (1)

[0059] Where, For the Pumping stations The neighborhood output tracking error of the outflow pool at time , For the final pump station Target water demand at all times, For the Pumping stations Target water level at the moment, No. Pumping stations Target water level at the moment, For the The error coefficient of each pumping station, For the The pumping station and The error coefficient of the interaction between the pumping stations, , is the total number of pumping stations in the cascade pumping station.

[0060] Compared with the prior art, the present invention no longer focuses on the tracking error of a single pump station, but introduces the neighborhood output tracking error (i.e. ) Solve the problem that some pumping stations cannot directly access the last-stage pumping station, making the scheduling of cascade pumping stations more accurate and improving the operating efficiency and economic benefits of cascade pumping stations.

[0061] Step 3: Identify each pump station Time input error of water pool and Output error of the water pool at any moment Time-varying parameters Estimated value of ,in , is the total number of pumping stations in the cascade pumping station.

[0062] The embodiment of the present invention calculates and analyzes the water volume characteristics of each pumping station to obtain a nonlinear data model between the water supply and water demand of each pumping station, where the input data is the water supply of each pumping station (corresponding to the water volume of the water inlet pool), and the output data is the water demand of each pumping station (corresponding to the water volume of the water outlet pool).

[0063] Specifically, the nonlinear data model of each pumping station is: .in, For the Pumping stations The target water level of the pool at all times, 、 Respectively represent Pumping stations The output and input at the moment, that is, The target water level of the outlet tank and the water level of the inlet tank of each pumping station; For the unknown nonlinear function of the pumping stations; 、 Respectively and of unknown order.

[0064] The embodiment of the present invention proposes the following assumptions for the above nonlinear data model:

[0065] Assumption 1: Nonlinear function about The partial derivatives of are continuous;

[0066] Assumption 2: When When , the above nonlinear mathematical model satisfies the generalized Lipschitz condition, namely: .in, for The output error of the water pool at any moment, , for Input error of water tank at all times, , is a positive constant.

[0067] The above assumption 1 means that the nonlinear function The partial derivatives of the variables are continuous, which is a common condition in controller design. Assumption 2 states that the rate of change of the output of each pumping station is limited by the rate of change of the input. If the change of the input of each pumping station is limited, then the change of its output must also be limited and cannot tend to infinity.

[0068] The above nonlinear data model dynamically satisfies Assumptions 1 and 2, and can be equivalently converted into the following dynamic linearized data model:

[0069] (2)

[0070] in, is a time-varying parameter that is bounded at any time, i.e. .

[0071] As all possible complex behavioral characteristics in the original nonlinear data model, it is mathematically difficult to obtain their exact true values, so it is necessary to design an estimation algorithm to estimate their numerical behavior.

[0072] The embodiment of the present invention determines the time-varying parameters Estimated value of As shown in formula (3).

[0073] (3)

[0074] Where, For the Pumping stations The estimated value of the time-varying parameter at time ; For the Pumping stations The estimated value of the time-varying parameter at time ; for The step size factor is used to control the estimation speed. Generally, for Normalization parameter to prevent the denominator from being zero. In general, ; for Input error of water tank at every moment; for Output error of the water pool at any moment.

[0075] The above formula (3) satisfies or ,but ,in yes The initial value of .

[0076] Assumption 3: For each pumping station and time step , there is an inequality (or ),in Represents a very small positive number. Without loss of generality, the present invention assumes .

[0077] Assumption 3 ensures that the output will not decrease as the corresponding input increases, reflecting the predictability and consistency of the input.

[0078] The present invention avoids the complexity of model-based control by establishing a nonlinear data model of input and output. The nonlinear data model is only used to generate input and output data and does not participate in the design and implementation of the corresponding controller in the subsequent step 4.

[0079] Step 4: Based on the estimated value of the time-varying parameter , neighborhood output tracking error and The water level of the inlet pool of each pumping station at any moment ,Sure The water level of the inlet pool of each pumping station at any moment .

[0080] The controller in step 4 of the embodiment of the present invention adopts a distributed model-free adaptive control algorithm to determine The water level of the inlet pool of each pumping station at any moment , specifically:

[0081] Step 4.1: Based on the estimated values ​​of time-varying parameters and neighborhood output tracking error Determine the increment value , specifically: Get the step size factor for updating the water level of each pump station inlet pool and regularization parameter ; According to the step factor , regularization parameter , the estimated values ​​of the time-varying parameters and neighborhood output tracking error Determine the increment value , as shown in formula (4).

[0082] (4)

[0083] in, For the pumping stations time The step size factor is used to make the control algorithm more general. ; For the pumping stations time The regularization parameter of ; For the Pumping stations The estimated value of the time-varying parameter at time ; For the Pumping stations The tracking error of the neighborhood output of the water pool at time t.

[0084] Step factor in the embodiment of the present invention According to Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations Determine, as shown in formula (5):

[0085] (5)

[0086] On this basis, for the Pumping stations exist , making As it approaches infinity, the system converges.

[0087] In the embodiment of the present invention, the regularization parameter According to the time-varying parameters Average value over time Sure.

[0088] In the embodiment of the present invention, the regularization parameter The update is as follows:

[0089] (6)

[0090] (7)

[0091] in, is a time-varying parameter Average value over time; is a time-varying parameter Average value over time; 、 is a positive definite constant.

[0092] Step 4.2: According to the increment value and The water level of the inlet pool of each pump station at any moment Sure The water level of the inlet pool of each pump station at any moment , as shown in formula (8):

[0093] (8)

[0094] The distributed model-free adaptive control algorithm in step 4 of the present invention dynamically adjusts the local control strategy based on the real-time sampled input and output data, which can realize the adaptive control of a single pumping station and can follow the target water demand of the final pumping station. Achieve desired control.

[0095] The data-driven cascade pump station scheduling optimization algorithm of the present invention combines the advantages of model-free adaptive control algorithm and multi-agent system, can achieve consistent control of cascade pump stations without relying on models, reduce the complexity of scheduling optimization, and have better robustness.

[0096] The second aspect of the present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned data-driven cascade pump station scheduling optimization method are implemented.

[0097] The scheduling optimization method of the present invention is simulated with a traditional model-free adaptive control algorithm to verify the effectiveness of the method of the present invention by comparing their convergence.

[0098] Consider a Figure 2 The communication topology diagram of the cascade pumping station shown in the figure can be converted into the following adjacency matrix:

[0099] , .

[0100] The embodiment of the present invention uses a typical nonlinear pump station dynamics equation as the input and output nonlinear data model:

[0101]

[0102] in, , This nonlinear data model only generates input and output data and does not participate in controller design.

[0103] The leader trajectory is:

[0104]

[0105] Total time step of the system , the controller parameters are:

[0106] ; ; ;

[0107] The system initialization parameters are:

[0108] ; ; ; ; ;

[0109] According to formula (4) The value range and Figure 2 The communication topology shown in the figure can be obtained in this example. The value range of is:

[0110] ; .

[0111] So one possibility that satisfies the conditions is:

[0112] ; .

[0113] In traditional model-free adaptive control algorithms, It needs to be set uniformly. In this example, it is set to 0.7.

[0114] According to formula (4) The update rule of , , initial value , while in the traditional model-free adaptive control algorithm, The global value is fixed as a constant, and is usually a small positive constant. In this example, Therefore, this example will also prove that Taking a larger value will not affect the convergence speed, because the system will update in real time , so that it adapts to a reasonable value.

[0115] like Figure 3 As shown, this is the water level coordinated control result of the traditional model-free adaptive control algorithm. Figure 4 This is the water level coordination result of the method of the present invention. It can be clearly seen that the method of the present invention can respond more quickly each time the leader changes, and the final water level converges to the desired water level more quickly. This proves the effectiveness and superiority of the present invention.

[0116] The data-driven cascade pump station scheduling optimization algorithm of the present invention optimizes the internal relationships of the cascade pump stations by improving the tracking error, and can ensure the efficient operation of the cascade pump stations and the full utilization of water resources.

[0117] The above descriptions are merely several embodiments of the present invention and do not constitute any form of limitation to the present invention. Although the present invention is disclosed as above in terms of preferred embodiments, they are not intended to limit the present invention. Any technician familiar with the present profession, without departing from the scope of the technical solution of the present invention, who makes slight changes or modifications using the technical contents disclosed above, is equivalent to an equivalent implementation case and falls within the scope of the technical solution.

Claims

1. A data-driven cascade pump station scheduling optimization method, characterized in that: include: Step 1. Preset The target water demand of the last pumping station at the moment and the target water level of the outlet pool of each pumping station in the cascade pumping station; Step 2: Determine the neighborhood output tracking error of each pump station outlet pool based on the target water demand and the target water level, specifically: Step 2.1, constructing a communication topology graph of the cascade pumping station and determining an adjacency matrix of the communication topology graph; Step 2.2, determining an error coefficient of each pumping station based on the adjacency matrix and the leadership relationship of the final pumping station to each of the remaining pumping stations; Step 2.3, determining a neighborhood output tracking error of each pump station outlet pool according to the error coefficient, the target water demand, and the target water level; Step 3: Identify each pump station The input error of the water tank at the moment is The estimated value of the time-varying parameter between the output error of the water tank at each moment; Step 4: Output tracking error according to the estimated value of the time-varying parameter, the neighborhood and The water level of each pump station inlet pool at the moment is determined The water level of the water inlet pool of each pumping station at all times.

2. The data-driven cascade pump station scheduling optimization method according to claim 1 is characterized in that: The error coefficient includes Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations , , is the total number of pumping stations in the cascade pumping station; then step 2.2 is specifically as follows: When When a pumping station directly accepts the leadership of the final pumping station, , ; otherwise , , is the element of the adjacency matrix.

3. The data-driven cascade pump station scheduling optimization method according to claim 2 is characterized in that: Step 2.3 is as follows: Determine the target water demand and The first difference of the target water levels of the pumping stations; Determine the The target water level of each pump station is the second difference of the target water levels of the pumping stations; Determine the first difference, the second difference and the error coefficient The tracking error of the neighborhood output of the water outlet pool of each pumping station.

4. The data-driven cascade pump station scheduling optimization method according to claim 3 is characterized in that: Determine the first difference, the second difference and the error coefficient The tracking error of the neighborhood output of the water outlet pool of each pump station is: Determine the first difference and The product of is recorded as the first value; Determine the second difference and The product of the second value is recorded, and the second value of the second value is determined by the product of the second value of the second value. The sum of the second values ​​of the pumping stations is recorded as the third value; Determine the first value based on the sum of the first value and the third value The tracking error of the neighborhood output of the water outlet pool of each pumping station.

5. The data-driven cascade pump station scheduling optimization method according to claim 2 is characterized in that: Step 4 is as follows: determining an incremental value according to an estimated value of the time-varying parameter and the neighborhood output tracking error; According to the increment value and The water level of each pump station inlet pool is determined at the moment The water level of the water inlet pool of each pumping station at all times.

6. The data-driven cascade pump station scheduling optimization method according to claim 5 is characterized in that: The incremental value is determined according to the estimated value of the time-varying parameter and the neighborhood output tracking error, specifically: Obtain the step size factor and regularization parameter for updating the water level of the inlet pool of each pumping station; An incremental value is determined according to the step size factor, the regularization parameter, the estimated value of the time-varying parameter, and the neighborhood output tracking error.

7. The data-driven cascade pump station scheduling optimization method according to claim 6 is characterized in that: The step factor is based on Error coefficient of each pumping station Hedi The pumping station and Error coefficient for interaction between pumping stations Sure.

8. The data-driven cascade pump station scheduling optimization method according to claim 6 is characterized in that: The regularization parameter is determined according to the average value of the time-varying parameter on the time axis.

9. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the data-driven cascade pump station scheduling optimization method as described in any one of claims 1 to 8 are implemented.

Images