Water pump turbine modeling method based on whale optimization algorithm and BP neural network

Abstract

The invention discloses a water pump turbine modeling method based on a whale optimization algorithm and a BP neural network. The water pump turbine modeling method comprises the following steps: S1, acquiring water pump turbine modeling original data by adopting an image reading tool; s2, unitizing modeling data of the water pump turbine; s3, processing modeling data of the water pump turbine based on improved Suter transformation; s4, determining a water pump turbine model structure based on the BP neural network; s5, optimizing hyper-parameters in the BP neural network-based pump turbine model by combining an Octave tool and a whale optimization algorithm; and S6, constructing a high-precision pump turbine model based on the BP neural network based on an Octave tool. According to the water pump type water turbine modeling method considering the water turbine characteristic curved surface and the model structure parameters, the influence of the water turbine characteristic curved surface and the model structure parameters can be fully considered, and an important model basis is provided for hydroelectric generating set optimization control research work.

Description

Technical Field

[0001] The present invention belongs to the technical field of safe and stable operation of hydropower units, and in particular relates to a pump turbine modeling method based on a whale optimization algorithm and a BP neural network. Background Art

[0002] As an important hydroelectric power generation equipment, pump-turbine has the key function of converting water energy into mechanical energy, and its performance directly affects the power generation efficiency and system reliability. Traditional turbine modeling methods mainly rely on physical equations and empirical formulas. Although they perform well in some simple cases, they have limitations in complex conditions such as nonlinear and dynamically changing conditions, making it difficult to accurately predict and optimize their performance.

[0003] In recent years, optimization algorithms have been increasingly used in engineering modeling and optimization. As a bionic intelligent algorithm, the Whale Optimization Algorithm (WOA) simulates the predation and chasing behavior of whale groups and can effectively perform global search and optimization. This makes WOA particularly suitable for dealing with complex nonlinear system modeling problems, such as performance modeling and optimization of pump-turbines.

[0004] Neural networks, especially back propagation neural networks (BP neural networks), are well-known for their ability to learn complex nonlinear relationships between data. BP neural networks are widely used in pattern recognition, data modeling, and prediction. Their ability lies in learning and extracting key patterns from large amounts of data, providing highly accurate prediction and analysis capabilities for complex systems.

[0005] Therefore, the present invention aims to solve the problem of insufficient accuracy under complex nonlinear and dynamically changing working conditions encountered in pump-turbine modeling. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a pump-turbine modeling method based on the whale optimization algorithm and the BP neural network. The high-precision nonlinear pump-turbine model obtained by the present invention can ensure the reliability of subsequent hydropower unit stability and optimization control research, thereby effectively ensuring the safe and stable operation of the hydropower unit.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0008] A pump-turbine modeling method based on whale optimization algorithm and BP neural network, the steps are:

[0009] S1, using the image reading tool to obtain the original data of pump-turbine modeling;

[0010] S2, normalize the original data of pump-turbine modeling;

[0011] S3, processing the raw data of pump-turbine modeling based on improved Suter transformation;

[0012] S4, determining the pump-turbine model structure based on BP neural network;

[0013] S5, combining Octave tool with whale optimization algorithm to optimize hyper parameters in pump-turbine model based on BP neural network;

[0014] S6, builds a high-precision pump-turbine model based on BP neural network based on Octave tool.

[0015] Preferably, the step S1 comprises the following steps:

[0016] S21, use the drawing reading tool to obtain the original data of pump-turbine modeling.

[0017] Ensuring that the original data required for pump-turbine modeling is obtained is the basis for accurately establishing a pump-turbine model, and is also the key basis for subsequently establishing and optimizing a pump-turbine model based on the whale optimization algorithm and BP neural network. The specific steps can be expanded as follows:

[0018] (1) Determine the data source:

[0019] Identify the available sources of data on pump-turbines, including design drawings, technical manuals, experimental data, etc. These data should include detailed information on the turbine's geometry, dimensional parameters, material properties, etc.

[0020] (2) Choose the appropriate image reading tool:

[0021] Choose a professional drawing reading tool suitable for reading and processing pump-turbine related data. Common tools include AutoCAD, GetData Graph Digitizer, etc., which can open and browse design drawings in standard CAD formats (such as DWG, DXF, etc.) or Fig formats and extract the required data.

[0022] (3) Import and browse design drawings:

[0023] Use the selected drawing reading tool to import the design drawings of the pump turbine. These drawings usually contain the turbine model, rated speed, runaway speed, data under normal operating conditions and runaway conditions (such as speed, guide vane opening, flow, output, etc.).

[0024] (4) Data extraction:

[0025] In the map reading tool, first establish an xy coordinate system, and then extract key data point by point. The acquired data should be divided into input parameters and output parameters. Input parameters should include: water head, flow, guide vane opening, etc.; output parameters should include: speed, output power, output, etc.

[0026] (5) Data recording and organization:

[0027] The extracted data are recorded in electronic documents and organized and classified according to modeling requirements.

[0028] Preferably, the step S2 is specifically:

[0029] According to the existing modeling data, the unit speed n is obtained after conversion. 11 , guide vane opening Y, unit flow Q 11 and unit moment M 11 Modeling dataset. 11 The acquisition of is shown in formula (1), Q 11 The acquisition of is shown in formula (2).

[0030]

[0031] Among them, γ = ρg, ρ is the density of water, g is the acceleration of gravity; D is the diameter of the impeller; H is the working water head; X is the speed; M 11 is the unit moment.

[0032] Preferably, the step S3 is specifically:

[0033] Processing pump-turbine modeling data based on the improved Suter transform involves a specific data processing method that aims to optimize the representation and feature extraction of data for more efficient subsequent modeling and analysis. The Suter transform is a signal processing technique used to improve the frequency domain analysis of data. It is often used to remove noise, highlight features, and enhance certain aspects of the signal. In pump-turbine modeling, the Suter transform can be used to highlight and analyze important features in the data. The following are the detailed steps and methods:

[0034] (1) Data preprocessing:

[0035] Before applying the Suter transform, the pump-turbine modeling data needs to be properly preprocessed, including data smoothing, denoising, or necessary filtering to ensure the quality and stability of the input data.

[0036] (2) Apply the improved Suter transform:

[0037] Definition: The variable x satisfies the transformation formula at time t:

[0038]

[0039] The relative value y of the guide vane opening at time t t As the independent variable, the numerical transformation is performed according to the improved Suter transformation formula as follows:

[0040]

[0041] Where n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; x is the custom variable after the improved Suter transformation. The value range of the auxiliary transformation parameters is: k2∈0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11max is the maximum unit moment, M 11r is the rated value of unit moment. y and C h The values ​​are 9, 0.8, 0.2 and 0.4 respectively.

[0042] Preferably, the step S4 is specifically:

[0043] Determining the pump-turbine model structure based on BP neural network involves determining the input layer and output layer of BP neural network. Specifically, the input layer of flow characteristic neural network selects x and y, and the output layer selects WH; the input layer of torque characteristic neural network selects x and y, and the output layer selects WM.

[0044] Preferably, the step S5 is specifically:

[0045] Octave tool and whale optimization algorithm are combined to optimize the hyperparameters in the pump-turbine model based on BP neural network, involving the number of hidden layer neurons, learning rate, and training times in the flow characteristic neural network model and the torque characteristic neural network model. In order to determine the number of hidden layer neurons, learning rate, and training times in the flow characteristic neural network model and the torque characteristic neural network model, the maximum N is first determined by formula (5). h , and then the mean square error MSE shown in formula (6) is used as the objective function of the whale optimization algorithm.

[0046]

[0047] Where N i is the number of neurons in the input layer, N ois the number of neurons in the output layer, Z is a constant between 0 and 10; k represents the number of sample data, Denotes the predicted data, D i Represents sample data.

[0048] Preferably, the step S6 is specifically:

[0049] A high-precision pump-turbine model based on BP neural network is constructed in Octave tool. Among them, the anti-Suter change of flow characteristic and torque characteristic neural network model should be considered. After obtaining WH(x,y) and WM(x,y), the relative working head and relative mechanical torque of the unit are solved according to formula (7). Furthermore, the actual flow of the turbine can be calculated according to the relative head.

[0050]

[0051] Where: x t is the value of variable x at time t, y t is the relative value y of the guide vane opening at time t, n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; h t+T is the h of the next sampling time T t ;m t+T is the m of the next sampling time T t ; The value range of auxiliary transformation parameters is: k2∈[0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11max is the maximum unit moment, M 11r It is the rated value of unit torque.

[0052] Preferably, the pump-turbine modeling method of the turbine characteristic surface and model structure parameters designed by the present invention includes three parts: obtaining pump-turbine modeling data, optimizing hyperparameters in the pump-turbine model based on BP neural network, and constructing a high-precision pump-turbine model based on BP neural network. The implementation steps mainly include: using a map reading tool to obtain the original data of pump-turbine modeling; unitizing the pump-turbine modeling data; processing the pump-turbine modeling data based on the improved Suter transformation; determining the pump-turbine model structure based on BP neural network; optimizing the hyperparameters in the pump-turbine model based on BP neural network by combining Octave tool and whale optimization algorithm; and constructing a high-precision pump-turbine model based on BP neural network based on Octave tool. The pump-turbine modeling method considering turbine characteristic surface and model structure parameters can fully consider the influence of turbine characteristic surface and model structure parameters, and provide an important model basis for further research on hydropower unit optimization control.

[0053] A pump-turbine modeling system based on a whale optimization algorithm and a BP neural network adopts the pump-turbine modeling method based on a whale optimization algorithm and a BP neural network, comprising:

[0054] Drawing reading tool: used to obtain the original data of pump-turbine modeling by using the drawing reading tool;

[0055] Data preprocessing module: used to unitize the original data of pump-turbine modeling;

[0056] Signal processing module: used to process the raw data of pump-turbine modeling based on improved Suter transform;

[0057] Machine learning module: used to determine the pump-turbine model structure based on BP neural network;

[0058] Optimization algorithm module: used to combine Octave tool with whale optimization algorithm to optimize the hyper parameters in the pump-turbine model based on BP neural network;

[0059] Simulation environment integration module: used to build a high-precision pump-turbine model based on BP neural network based on Octave tool.

[0060] A computer device comprising:

[0061] one or more processors;

[0062] The processor is used to store one or more programs;

[0063] When the one or more programs are executed by the one or more processors, a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network is implemented.

[0064] A computer-readable storage medium stores a computer program, which, when executed, implements a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network as described above.

[0065] The present invention can achieve the following beneficial effects:

[0066] 1. The present invention combines the advantages of the whale optimization algorithm and the BP neural network to propose an innovative pump-turbine modeling method. By optimizing the parameters and structure of the neural network through WOA, the performance and convergence speed of the neural network are improved. This combination can not only effectively overcome the limitations of traditional methods, such as avoiding local optimality and improving model accuracy, but also better adapt to the complex characteristics of pump-turbines under different operating conditions.

[0067] 2. The present invention provides a new and efficient pump-turbine modeling method by introducing the whale optimization algorithm to optimize the structure and parameters of the neural network. This method not only improves the accuracy and efficiency of turbine modeling in theory, but also has broad practical application prospects, including but not limited to the design optimization of hydropower generation systems, energy efficiency improvement, and enhanced system operation safety.

[0068] 3. The high-precision nonlinear pump-turbine model obtained by the present invention can ensure the reliability of subsequent hydropower unit stability and optimization control research, thereby effectively ensuring the safe and stable operation of the hydropower unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0070] Figure 1 Designing steps for the method of the present invention;

[0071] Figure 2 The calculation process of the whale optimization algorithm used in the present invention;

[0072] Figure 3 This is a schematic diagram of a nonlinear pump-turbine model based on a BP neural network constructed in the present invention. DETAILED DESCRIPTION

[0073] The preferred solution is Figures 1 to 3 As shown, a pump-turbine modeling method based on whale optimization algorithm and BP neural network includes the following steps:

[0074] (1) Using the drawing reading tool to obtain the original data of pump-turbine modeling

[0075] Ensuring that the original data required for pump-turbine modeling is obtained is the basis for accurately establishing a pump-turbine model, and is also the key basis for subsequently establishing and optimizing a pump-turbine model based on the whale optimization algorithm and BP neural network. The specific steps can be expanded as follows:

[0076] 1) Determine the data source:

[0077] Identify the available sources of data on pump-turbines, including design drawings, technical manuals, experimental data, etc. These data should include detailed information on the turbine's geometry, dimensional parameters, material properties, etc.

[0078] 2) Choose the appropriate image reading tool:

[0079] Choose a professional drawing reading tool suitable for reading and processing pump-turbine related data. Common tools include AutoCAD, GetData Graph Digitizer, etc., which can open and browse design drawings in standard CAD formats (such as DWG, DXF, etc.) or Fig formats and extract the required data.

[0080] 3) Import and browse design drawings:

[0081] Use the selected drawing reading tool to import the design drawings of the pump turbine. These drawings usually contain the turbine model, rated speed, runaway speed, data under normal operating conditions and runaway conditions (such as speed, guide vane opening, flow, output, etc.).

[0082] 4) Data extraction:

[0083] In the map reading tool, first establish an xy coordinate system, and then extract key data point by point. The acquired data should be divided into input parameters and output parameters. Input parameters should include: water head, flow, guide vane opening, etc.; output parameters should include: speed, output power, output, etc.

[0084] 5) Data recording and organization:

[0085] The extracted data are recorded in electronic documents and organized and classified according to modeling requirements.

[0086] (2) Unitize the pump-turbine modeling data

[0087] According to the existing modeling data, the unit speed n is obtained after conversion. 11 , guide vane opening Y, unit flow Q 11 and unit moment M 11 Modeling dataset. 11 The acquisition of is shown in formula (8), Q 11The acquisition of is shown in formula (9).

[0088]

[0089] Among them, γ = ρg, ρ is the density of water, g is the acceleration of gravity; D is the diameter of the impeller; H is the working water head; X is the speed; M 11 is the unit moment.

[0090] (3) Processing pump-turbine modeling data based on improved Suter transformation

[0091] Processing pump-turbine modeling data based on the improved Suter transform involves a specific data processing method that aims to optimize the representation and feature extraction of data for more efficient subsequent modeling and analysis. The Suter transform is a signal processing technique used to improve the frequency domain analysis of data. It is often used to remove noise, highlight features, and enhance certain aspects of the signal. In pump-turbine modeling, the Suter transform can be used to highlight and analyze important features in the data. The following are the detailed steps and methods:

[0092] 1) Data preprocessing:

[0093] Before applying the Suter transform, the pump-turbine modeling data needs to be properly preprocessed, including data smoothing, denoising, or necessary filtering to ensure the quality and stability of the input data.

[0094] 2) Apply the improved Suter transform:

[0095] Definition: The variable x satisfies the transformation formula at time t:

[0096]

[0097] The relative value y of the guide vane opening at time t t As the independent variable, the numerical transformation is performed according to the improved Suter transformation formula as follows:

[0098]

[0099] Where n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; x is the custom variable after the improved Suter transformation. The value range of the auxiliary transformation parameters is: k2∈[0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11maxis the maximum unit moment, M 11r is the rated value of unit moment. y and C h The values ​​are 9, 0.8, 0.2 and 0.4 respectively.

[0100] (4) Determine the pump-turbine model structure based on BP neural network

[0101] Determining the pump-turbine model structure based on BP neural network involves determining the input layer and output layer of BP neural network. Specifically, the input layer of flow characteristic neural network selects x and y, and the output layer selects WH; the input layer of torque characteristic neural network selects x and y, and the output layer selects WM.

[0102] (5) Combining Octave tool with Whale Optimization Algorithm to Optimize Hyperparameters in Pump-Turbine Model Based on BP Neural Network

[0103] The Octave tool and the whale optimization algorithm are combined to optimize the hyperparameters in the pump-turbine model based on the BP neural network, involving the number of hidden layer neurons, learning rate, and training times in the flow characteristic neural network model and the torque characteristic neural network model. In order to determine the number of hidden layer neurons, learning rate, and training times in the flow characteristic neural network model and the torque characteristic neural network model, the maximum N is first determined by formula (12): h , and then the mean square error MSE shown in formula (13) is used as the objective function of the whale optimization algorithm.

[0104]

[0105]

[0106] Where N i is the number of neurons in the input layer, N o is the number of neurons in the output layer, Z is a constant between 0 and 10; k represents the number of sample data, Denotes the predicted data, D i Represents sample data.

[0107] Among them, the Whale Optimization Algorithm (WOA) is an optimization algorithm based on bionic intelligence, which simulates the predation behavior and social behavior characteristics of whale groups. The algorithm achieves efficient solution capabilities for optimization problems by simulating the predation and chasing behaviors of whales. In the Whale Optimization Algorithm, each whale is represented as a candidate solution for a solution, and the chasing behavior of whales reflects the balance between exploration and utilization in the search process. Specifically, the algorithm optimizes the quality of the solution by updating the position of each whale (i.e., the parameter value of the solution), and at the same time simulates the ability of global search and local search by using different predation behaviors, so that the algorithm can effectively explore and optimize in the solution space.

[0108] The core features of the whale optimization algorithm include the following:

[0109] 1) Global search capability: By simulating the migration behavior of whales, the algorithm can explore extensively in the solution space, which helps to find the global optimal solution.

[0110] 2) Local search capability: After discovering a potential solution, the algorithm simulates the hunting behavior of whales to conduct a refined local search, which helps to optimize the accuracy and convergence speed of the solution.

[0111] 3) Simplicity and efficiency: The algorithm has a simple implementation and efficient solving ability, and is suitable for solving a variety of optimization problems.

[0112] The WOA algorithm simulates the whale's hunting behavior, which mainly includes three stages: surrounding prey, attacking with a bubble net, and searching for prey. The formula for the surrounding prey stage is as follows:

[0113]

[0114] Where: t represents the number of iterations; t max Indicates the maximum number of iterations; is the coefficient vector; A random number between the range [0, 1]; represents the search particle, Represents the target prey, which is the position vector of the current optimal solution, and is continuously updated in iterations to obtain a better solution; is the convergence factor that decreases linearly from 2 to 0 during the iteration process. The WOA bubble attack behavior is shown in equations (19) and (20), where p is a random number between [0, 1].

[0115]

[0116] Where: b is the shape parameter of the logarithmic spiral, l is a random number between [-1, 1], Represents the distance between the search particle and the current optimal solution.

[0117] The mathematical model of the whale's search for prey is:

[0118]

[0119] In the formula, represents a search particle randomly selected from the population.

[0120] The WOA algorithm first randomly initializes a set of parameters and uses the fitness function to calculate and record the current best solution. Subsequently, the algorithm adjusts and updates these parameters according to a specific search mechanism and continues to iterate until the preset maximum number of iterations is reached. Finally, the WOA algorithm stops running and outputs the global optimal solution found.

[0121] (6) Constructing a high-precision pump-turbine model based on BP neural network using Octave tool

[0122] A high-precision pump-turbine model based on BP neural network is constructed in Octave tool. Among them, the anti-Suter change of flow characteristic and torque characteristic neural network model should be considered. After obtaining WH(x,y) and WM(x,y), the relative working head and relative mechanical torque of the unit are solved according to formula (14). Furthermore, the actual flow of the turbine can be calculated according to the relative head.

[0123]

[0124] Where: x t is the value of variable x at time t, y t is the relative value y of the guide vane opening at time t, n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; h t+T is the h of the next sampling time T t ;m t+T is the m of the next sampling time T t ; The value range of auxiliary transformation parameters is: k2∈[0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11max is the maximum unit moment, M 11r It is the rated value of unit torque.

[0125] The pump-turbine modeling method designed by the present invention taking into account the optimal parameters of the BP neural network can fully consider the influence of the parameters and structure of the neural network on the accuracy of the pump-turbine model, and effectively ensure the safe and stable operation of the hydropower unit.

[0126] As the core equipment for power generation in hydropower stations, the safe and stable operation of hydropower units has become a hot topic in the industry. Aiming at the limitations of traditional pump-turbine modeling methods, the present invention proposes a pump-turbine modeling method that combines a whale optimization algorithm with a BP neural network.

[0127] A computer device comprising:

[0128] one or more processors;

[0129] The processor is used to store one or more programs;

[0130] When the one or more programs are executed by the one or more processors, a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network is implemented.

[0131] A computer-readable storage medium stores a computer program, which, when executed, implements a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network as described above.

[0132] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A pump-turbine modeling method based on whale optimization algorithm and BP neural network, characterized in that The following steps are involved: S1, using the image reading tool to obtain the original data of pump-turbine modeling; S2, normalize the original data of pump-turbine modeling; S3, processing the raw data of pump-turbine modeling based on improved Suter transformation; S4, determining the pump-turbine model structure based on BP neural network; S5, combining Octave tool with whale optimization algorithm to optimize hyper parameters in pump-turbine model based on BP neural network; S6, builds a high-precision pump-turbine model based on BP neural network based on Octave tool.

2. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S1 comprises the following steps: (1) Determine the data source: Confirm the available sources of data on pump-turbines, including design drawings, technical manuals and experimental data, which at least include parameter information such as the geometry, dimensional parameters and material properties of the turbine; (2) Choose the appropriate image reading tool: Select and use a drawing reading tool suitable for reading and processing pump-turbine related data, such as AutoCAD and GetData Graph Digitizer; the drawing reading tool should be able to open and browse the design drawings in standard CAD format or Fig format and extract the required data; (3) Import and browse design drawings: Use the selected drawing reading tool to import the design drawings of the pump-turbine, which include the turbine model, rated speed, runaway speed, data under normal operating conditions and runaway conditions; (4) Data extraction: In the map reading tool, first establish an xy coordinate system, then extract key data point by point, and divide the acquired data into input parameters and output parameters; Input parameters should include: water head, flow rate, guide vane opening; Output parameters include: speed, output power, output; (5) Data recording and organization: The extracted data are recorded in electronic documents and organized and classified according to modeling requirements.

3. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S2 is specifically as follows: According to the existing modeling original data, the unit speed n is obtained after conversion. 11 , guide vane opening Y, unit flow Q 11 and unit moment M 11 Modeling dataset; n 11 The acquisition of is shown in formula (1), Q 11 The acquisition of is shown in formula (2): Among them, γ = ρg, ρ is the density of water, g is the acceleration of gravity; D is the diameter of the impeller; H is the working water head; X is the speed; M 11 is the unit moment.

4. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S3 is specifically as follows: (1) Data preprocessing: Before applying the Suter transformation, the pump-turbine modeling data is preprocessed, including data smoothing, denoising or filtering to ensure the quality and stability of the input data; (2) Apply the improved Suter transform: Definition: The variable x satisfies the transformation formula at time t: The relative value y of the guide vane opening at time t t As the independent variable, the numerical transformation is performed according to the improved Suter transformation formula as follows: Where n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; x is the custom variable after the improved Suter transformation; the value range of the auxiliary transformation parameter is: k2∈[0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11max is the maximum unit moment, M 11r It is the rated value of unit torque.

5. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S4 is specifically as follows: Determine the input layer and output layer of the BP neural network: select x and y for the flow characteristic neural network input layer, and select WH for the output layer; select x and y for the torque characteristic neural network input layer, and select WM for the output layer.

6. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S5 is specifically as follows: Determine the number of hidden layer neurons, learning rate, and training times in the flow characteristic neural network model and the torque characteristic neural network model. First, determine the maximum N by formula (5): h , and then the mean square error MSE shown in formula (6) is used as the objective function of the whale optimization algorithm: Where N i is the number of neurons in the input layer, N o is the number of neurons in the output layer, Z is a constant between 0 and 10; k represents the number of sample data, Denotes the predicted data, D i Represents sample data.

7. The pump-turbine modeling method based on whale optimization algorithm and BP neural network according to claim 1 is characterized in that: The step S6 is specifically as follows: Considering the inverse Suter change of the flow characteristic and torque characteristic neural network model, after obtaining WH(x, y) and WM(x, y), the relative working water head and relative mechanical torque of the unit are solved according to formula (7); the actual flow of the turbine is calculated according to the relative water head: Where: x t is the value of variable x at time t, y t is the relative value y of the guide vane opening at time t, n t ,q t 、h t 、m t are the relative values ​​of speed, flow, head and torque at time t respectively; h t+T is the h of the next sampling time T t ;m t+T is the m of the next sampling time T t ; The value range of auxiliary transformation parameters is: k2∈[0.5,1.2],C y ∈[0.1,0.3],C h ∈[0.4,0.6],M 11max is the maximum unit moment, M 11r It is the rated value of unit torque.

8. A pump-turbine modeling system based on a whale optimization algorithm and a BP neural network, which adopts a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network according to any one of claims 1 to 7, comprising: Drawing reading tool: used to obtain the original data of pump-turbine modeling by using the drawing reading tool; Data preprocessing module: used to unitize the original data of pump-turbine modeling; Signal processing module: used to process the raw data of pump-turbine modeling based on improved Suter transform; Machine learning module: used to determine the pump-turbine model structure based on BP neural network; Optimization algorithm module: used to combine Octave tool with whale optimization algorithm to optimize the hyper parameters in the pump-turbine model based on BP neural network; Simulation environment integration module: used to build a high-precision pump-turbine model based on BP neural network based on Octave tool.

9. A computer device, characterized in that: include: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a pump-turbine modeling method based on a whale optimization algorithm and a BP neural network as described in any one of claims 1 to 7 is implemented.

Images