Hydroelectric power generation prediction method and system based on VMD-LSTM-KAN model

Abstract

The invention relates to a hydroelectric power generation prediction method and system based on a VMD-LSTM-KAN model. The method comprises the following steps that historical power generation data and historical meteorological data of a target hydropower station are collected and preprocessed; carrying out VMD variational mode decomposition on the preprocessed historical power generation data, and decomposing the historical power generation data into a plurality of mode components from time sequence data; combining the plurality of modal components and historical power generation data into a multi-dimensional output matrix; fusing the preprocessed historical meteorological data with the multi-dimensional output matrix to generate an input data set; constructing a prediction model based on the LSTM network, performing nonlinear optimization on an output layer of the LSTM network in combination with a KAN network, and training the constructed prediction model through an input data set; and predicting a modal component prediction value and a power generation prediction value of the target hydropower station in a future time period through the trained prediction model.

Description

Technical Field

[0001] The present invention relates to a hydropower generation prediction method and system based on a VMD-LSTM-KAN model, belonging to the technical field of power generation prediction. Background Art

[0002] Hydropower station power generation forecasting is a crucial link in the operation of the power system. Its accuracy is directly related to the efficiency of power dispatching, the rational use of energy, and the economy and stability of system operation. However, hydropower station power generation is affected by a variety of complex factors, which increases the difficulty of prediction. The dynamic changes in meteorological conditions are key factors affecting power generation, including relative humidity, air temperature, rainfall and snowfall, which have significant uncertainty and nonlinear characteristics. Especially in the case of drastic short-term weather changes, the coupling between meteorological factors makes it difficult to model the fluctuation characteristics of power generation through simple linear relationships. At the same time, the power generation process of hydropower stations is also affected by a variety of internal and external factors such as water inflow, reservoir scheduling rules and power generation equipment efficiency. This multi-factor, multi-level coupling relationship further highlights the complexity of the prediction task.

[0003] Traditional time series forecasting methods, such as the autoregressive integrated moving average model (ARIMA) and support vector machine (SVM), although they have certain advantages in dealing with linear time series problems, have obvious shortcomings in dealing with nonlinear characteristics, high-frequency noise, and dynamic meteorological changes. These methods usually rely on linear assumptions or fixed kernel function mappings, and it is difficult to effectively capture nonlinear relationships in complex systems. In particular, these methods have poor robustness when dealing with high-frequency noise and outliers that are common in hydropower station power generation data. In addition, there are often missing values ​​in the historical data and meteorological data of hydropower stations due to equipment failures, collection errors, etc. Traditional methods are difficult to adapt to the incompleteness of these data, and the prediction results often have large deviations.

[0004] In recent years, with the rapid development of artificial intelligence technology, machine learning and deep learning methods have been widely used in the field of time series prediction, especially models such as long short-term memory network (LSTM) and convolutional neural network (CNN) have attracted much attention due to their powerful feature extraction capabilities. These models have significant advantages in capturing the long-term dependencies and nonlinear dynamic characteristics of time series data. However, a single deep learning model still faces many challenges. For example, LSTM has limited ability to deal with high-frequency noise and complex nonlinear relationships. Although CNN performs well in local feature extraction, it does not adequately model the dynamic characteristics of time series. In addition, existing deep learning models generally lack the ability to systematically process multi-frequency characteristics, making it difficult to effectively separate low-frequency trends, medium-frequency features, and high-frequency noise, which makes it difficult to capture the overall characteristics of power generation data in a comprehensive manner. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention proposes a hydropower generation prediction method and system based on the VMD-LSTM-KAN model.

[0006] The technical solution of the present invention is as follows:

[0007] On the one hand, the present invention proposes a hydropower generation prediction method based on the VMD-LSTM-KAN model, comprising the following steps:

[0008] Collect historical power generation data and historical meteorological data of the target hydropower station and perform preprocessing;

[0009] Perform VMD variational modal decomposition on the preprocessed historical power generation data to decompose the historical power generation data from time series data into multiple modal components;

[0010] Combining multiple modal components with historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents a feature of a corresponding modal component or historical power generation data;

[0011] The preprocessed historical meteorological data is fused with the multi-dimensional output matrix to generate an input data set;

[0012] Build a prediction model based on LSTM network, and combine KAN network with the output layer of LSTM network for nonlinear optimization, and train the constructed prediction model through input data set;

[0013] The trained prediction model is used to predict the modal component prediction value and power generation prediction value of the target hydropower station in the future period.

[0014] As a preferred implementation, the step of collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them includes:

[0015] Obtain historical power generation data and use the spline interpolation method to fill in the missing data;

[0016] Obtain historical meteorological data, including relative humidity data, rain water content data, snow water content data and air temperature;

[0017] The historical power generation data and historical meteorological data are normalized.

[0018] As a preferred implementation, in the step of performing VMD variational mode decomposition on the preprocessed historical power generation data:

[0019] By setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data.

[0020] As a preferred implementation, the method of combining the KAN network with the output layer of the LSTM network for nonlinear optimization is specifically as follows:

[0021] The output of the LSTM network is received through the KAN network, and the output of the LSTM network is nonlinearly optimized using the following formula:

[0022]

[0023] Among them, h t is the output of the LSTM network, S(x, c i ) is the basis spline function, ω i represents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

[0024] On the other hand, the present invention also provides a hydropower generation prediction system based on the VMD-LSTM-KAN model, comprising:

[0025] Data acquisition module, used to collect historical power generation data and historical meteorological data of the target hydropower station and perform preprocessing;

[0026] A data processing module is used to perform VMD variational modal decomposition on the pre-processed historical power generation data, and decompose the historical power generation data from time series data into multiple modal components;

[0027] A feature transformation module, used to combine multiple modal components and historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents a feature of a corresponding modal component or historical power generation data;

[0028] The data set construction module is used to fuse the pre-processed historical meteorological data with the multi-dimensional output matrix to generate the input data set;

[0029] The model training module is used to build a prediction model based on the LSTM network, and combine the KAN network to perform nonlinear optimization on the output layer of the LSTM network, and train the built prediction model through the input data set;

[0030] The prediction module is used to predict the modal component prediction value and power generation prediction value of the target hydropower station in the future period through the trained prediction model.

[0031] As a preferred implementation, the step of collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them includes:

[0032] Obtain historical power generation data and use the spline interpolation method to fill in the missing data;

[0033] Obtain historical meteorological data, including relative humidity data, rain water content data, snow water content data and air temperature;

[0034] The historical power generation data and historical meteorological data are normalized.

[0035] As a preferred implementation, in the step of performing VMD variational mode decomposition on the preprocessed historical power generation data:

[0036] By setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data.

[0037] As a preferred implementation, the method of combining the KAN network with the output layer of the LSTM network for nonlinear optimization is specifically as follows:

[0038] The output of the LSTM network is received through the KAN network, and the output of the LSTM network is nonlinearly optimized using the following formula:

[0039]

[0040] Among them, h t is the output of the LSTM network, S(x, c i ) is the basis spline function, ω i represents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

[0041] On the other hand, the present invention also proposes an electronic 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 program, the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any embodiment of the present invention is implemented.

[0042] On the other hand, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any embodiment of the present invention.

[0043] Additional aspects and advantages of the present invention will be set forth in the following description, and some of them will be apparent from the description, or may be understood by practicing the present invention. In addition, the various aspects and advantages of the present invention may be realized and obtained by the method steps and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1This is a flow chart of a method according to Embodiment 1 of the present invention;

[0045] Figure 2 This is a schematic diagram of the workflow of Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0048] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0049] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0050] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0051] Embodiment 1:

[0052] Due to the characteristics of hydropower station power generation data, the prediction method needs to comprehensively consider multi-frequency decomposition, nonlinear modeling and time series dependency modeling capabilities. This type of data contains both low-frequency trends (such as seasonal changes), medium-frequency characteristics (such as periodic fluctuations) and high-frequency noise (such as anomalies and short-term fluctuations). Existing prediction methods are often difficult to fully capture these characteristics, resulting in prediction accuracy and robustness that are difficult to meet actual needs. Therefore, this embodiment provides a new prediction method that can combine multi-frequency decomposition technology, nonlinear enhancement mechanism and time series relationship modeling to cope with the complexity and diversity of hydropower station power generation prediction tasks and meet the actual application needs of power system scheduling and resource management.

[0053] Therefore, the present embodiment provides a hydropower generation prediction method based on the VMD-LSTM-KAN model, which uses variational mode decomposition (VMD) to extract the multi-frequency characteristics of hydropower generation data, captures the temporal dependency through the long short-term memory network (LSTM), and combines the Kolmogorov-Arnold network (KAN) to enhance the nonlinear modeling capability. By fusing meteorological data with the decomposed mode, the present invention achieves high-precision prediction of complex hydropower generation systems with good robustness and adaptability.

[0054] See Figure 1 and Figure 2 The method of this embodiment specifically includes the following steps:

[0055] S100, collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them. The historical meteorological data is obtained through the data of the European Center for Medium-Range Weather Forecasts (ECMWF).

[0056] In this embodiment, for historical power generation data, the spline interpolation method is used to fill the missing values, and the integrity and continuity of the data are ensured by constructing a smoothing function. The formula of the spline interpolation method is as follows:

[0057] S(x)=a k (xx k ) 3 +b k (xx k ) 2 +c k (xx k )+d k ,x∈[x k ,x k+1 ];

[0058] In the above formula, x is the original historical power generation data, S(x) is the interpolation function, and a k , b k , c k , d k is the preset interpolation coefficient. This method can ensure that the interpolated data maintains the same trend characteristics as the original data and avoids the introduction of abnormal deviations.

[0059] For meteorological data, including relative humidity, rainfall, snowfall, and air temperature, the data are normalized to the range of [0,1], and the specific formula is:

[0060]

[0061] In the above formula, x is the original meteorological data, x min and x maxare the minimum and maximum values ​​of the data set, respectively. Normalization eliminates the dimensional differences between different features, enhances the consistency of data and the stability of model training. The normalized data will be used as external feature input to supplement the environmental information in the power generation data. The external feature input matrix X is as follows:

[0062] X=[Humidity, RainwaterContent, SnowwaterContent, Temperature];

[0063] In the above formula, Humidity, RainwaterContent, SnowwaterContent, and Temperature represent the normalized humidity, rainfall, snowfall, and air temperature data, respectively.

[0064] The normalized meteorological data combined with the historical power generation data of the hydropower station can enable the model to uniformly process feature data from different sources during the subsequent prediction model training process, thereby improving the robustness and efficiency of model training.

[0065] S200, performing VMD variational modal decomposition on the preprocessed historical power generation data, decomposing the historical power generation data from time series data into multiple modal components; VMD variational modal decomposition can decompose complex time series signals into several modal components (IMFs) by optimizing the objective function, and each modal component represents a specific frequency bandwidth.

[0066] In this embodiment, the objective function of VMD variational mode decomposition is set as:

[0067]

[0068] In the above formula, min represents the minimization operation, u k is the kth modal component, ω k is the center frequency of the modal component, K is the preset number of modes, * represents the convolution operation, δ(t) is the impulse function, and j is an imaginary number.

[0069] In this embodiment, by setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data. In the VMD decomposition process, the number of modes, bandwidth constraint parameters and convergence conditions are set to ensure that the decomposition results are both accurate and have high computational efficiency. The low-frequency mode captures the long-term trend of the data, the medium-frequency mode reflects seasonal and cyclical fluctuations, and the high-frequency mode extracts short-term fluctuations and noise characteristics.

[0070] S300, combining multiple modal components and historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents a feature of a corresponding modal component or historical power generation data.

[0071] In this embodiment, the multi-dimensional output matrix Y is as follows:

[0072] Y=[IMF1,IMF2,...,IMF K ,Original Data];

[0073] In the above formula, IMF K represents the Kth modal component, and Original Data represents the historical power generation data.

[0074] S400, fusing the pre-processed historical meteorological data with the multi-dimensional output matrix, that is, fusing the external feature input matrix X and the multi-dimensional output matrix Y to generate an input data set.

[0075] S500, construct a prediction model based on the LSTM network, and combine the KAN network to perform nonlinear optimization on the output layer of the LSTM network, and train the constructed prediction model through the input data set.

[0076] Among them, the long short-term memory network (LSTM) models the feature matrix in time series and extracts key temporal dynamic features. The memory unit and gating mechanism of LSTM are composed of forget gate, input gate and output gate respectively. The specific process is as follows:

[0077] (1) Deciding to discard information

[0078] The first step in the LSTM network is to decide what information to discard from the cell state. This decision is made through a gate called the "forget gate". This gate reads ht-1 and xt and outputs a value between 0 and 1 for each number in the cell state Ct-1. 1 means "totally keep" and 0 means "totally discard".

[0079] f t =σ(W f ×[h t-1 ,x t ]+b f );

[0080] (2) Determine the updated information

[0081] The next step is to decide what new information to store in the cell state. There are two parts to this. First, a sigmoid layer called the “input gate” decides what value we are going to update. Then, a tanh layer creates a new candidate value vector Add to status.

[0082] i t =σ(W i ×[h t-1 ,x t ]+b i );

[0083]

[0084] (3) Update cell status

[0085] Update the old cell state, Ct-1 is updated to Ct, and the state is updated based on the information calculated in the previous step. t Multiply, discard the information we are sure we need to discard, and then add is the new candidate value.

[0086]

[0087] (4) Output information

[0088] We first run a sigmoid layer to determine which part of the cell state will be output. Next, the cell state is processed through tanh (to get a value between -1 and 1) and multiplied with the output of the sigmoid gate. Finally, the network will output only the part of the information we determined to output based on the cell state.

[0089] o t =σ(W o ×[h t-1 ,x t ]+b o );

[0090] h t =o t ×tanh(C t );

[0091] Through the modeling capabilities of LSTM, the model can capture the time series patterns in the input features and provide input for subsequent nonlinear enhancement modeling.

[0092] KAN network (Kolmogorov-Arnold) outputs the feature vector h of LSTM t Perform nonlinear optimization. KAN uses a spline activation function to achieve nonlinear enhancement, specifically:

[0093]

[0094] Among them, h t is the output of the LSTM network, S(x, c i ) is the basis spline function, ω irepresents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

[0095] The KAN network uses a set of learnable control points and dynamic adjustment of spline functions to achieve nonlinear enhancement. The number of control points is set through experimental optimization to balance the nonlinear expression ability and computational complexity of the model. During the training process, the position and weight of the control points are optimized through the back-propagation algorithm.

[0096] The predicted values ​​​​of the model's final output are as follows:

[0097] y pred =W out ·h t '+b out

[0098] Among them, h t ' is the LSTM output after KAN enhancement, W out and b out are weight and bias parameters, y pred Represents the forecast value of hydroelectric power generation.

[0099] Based on the above LSTM-KAN network, the weight parameters of the KAN network are optimized through back propagation during the training process to ensure that the spline function can flexibly adjust the nonlinear mapping relationship between input and output. The prediction results are evaluated by indicators such as mean square error (MSE), root mean square error (RMSE) and mean absolute error (MAE).

[0100] S600, predicting modal component prediction values ​​and power generation prediction values ​​of the target hydropower station in the future period through the trained prediction model.

[0101] Embodiment 2:

[0102] This embodiment provides a hydropower generation prediction system based on a VMD-LSTM-KAN model, including:

[0103] A data acquisition module, used to collect historical power generation data and historical meteorological data of the target hydropower station and perform preprocessing; this module is used to implement the function of step S100 in the first embodiment, which will not be described in detail here;

[0104] A data processing module, used to perform VMD variational modal decomposition on the preprocessed historical power generation data, and decompose the historical power generation data from time series data into multiple modal components; this module is used to implement the function of step S200 in the first embodiment, and will not be described in detail here;

[0105] A feature transformation module, used to combine multiple modal components and historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents the feature of the corresponding modal component or historical power generation data; this module is used to implement the function of step S300 in the first embodiment, and will not be repeated here;

[0106] A data set construction module, used to fuse the pre-processed historical meteorological data with the multi-dimensional output matrix to generate an input data set; this module is used to implement the function of step S400 in the first embodiment, which will not be described in detail here;

[0107] A model training module is used to construct a prediction model based on an LSTM network, and to perform nonlinear optimization on the output layer of the LSTM network in combination with a KAN network, and to train the constructed prediction model through an input data set; this module is used to implement the function of step S500 in the first embodiment, and will not be described in detail here;

[0108] The prediction module is used to predict the modal component prediction value and power generation prediction value of the target hydropower station in the future period through the trained prediction model; this module is used to implement the function of step S600 in embodiment 1, which will not be repeated here.

[0109] As a preferred implementation of this embodiment, the step of collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them includes:

[0110] Obtain historical power generation data and use the spline interpolation method to fill in the missing data;

[0111] Obtain historical meteorological data, including relative humidity data, rain water content data, snow water content data and air temperature;

[0112] The historical power generation data and historical meteorological data are normalized.

[0113] As a preferred implementation of this embodiment, in the step of performing VMD variational mode decomposition on the preprocessed historical power generation data:

[0114] By setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data.

[0115] As a preferred implementation of this embodiment, the method of combining the KAN network with the output layer of the LSTM network for nonlinear optimization is specifically as follows:

[0116] The output of the LSTM network is received through the KAN network, and the output of the LSTM network is nonlinearly optimized using the following formula:

[0117]

[0118] Among them, h t is the output of the LSTM network, S(x, c i ) is the basis spline function, ω i represents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

[0119] Embodiment three:

[0120] This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any embodiment of the present invention is implemented.

[0121] Embodiment 4:

[0122] This embodiment proposes a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any embodiment of the present invention is implemented.

[0123] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0124] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0125] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0126] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0127] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A hydropower generation prediction method based on VMD-LSTM-KAN model, characterized in that: The following steps are involved: Collect historical power generation data and historical meteorological data of the target hydropower station and perform preprocessing; Perform VMD variational modal decomposition on the preprocessed historical power generation data to decompose the historical power generation data from time series data into multiple modal components; Combining multiple modal components with historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents a feature of a corresponding modal component or historical power generation data; The preprocessed historical meteorological data is fused with the multi-dimensional output matrix to generate an input data set; Build a prediction model based on LSTM network, and combine KAN network with the output layer of LSTM network for nonlinear optimization, and train the constructed prediction model through input data set; The trained prediction model is used to predict the modal component prediction value and power generation prediction value of the target hydropower station in the future period.

2. The hydropower generation prediction method based on the VMD-LSTM-KAN model according to claim 1 is characterized in that: The step of collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them includes: Obtain historical power generation data and use the spline interpolation method to fill in the missing data; Obtain historical meteorological data, including relative humidity data, rain water content data, snow water content data and air temperature; The historical power generation data and historical meteorological data are normalized.

3. The hydropower generation prediction method based on the VMD-LSTM-KAN model according to claim 1 is characterized in that: In the step of performing VMD variational mode decomposition on the preprocessed historical power generation data: By setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data.

4. The hydropower generation prediction method based on the VMD-LSTM-KAN model according to claim 1 is characterized in that: The method of combining the KAN network with the output layer of the LSTM network for nonlinear optimization is specifically as follows: The output of the LSTM network is received through the KAN network, and the output of the LSTM network is nonlinearly optimized using the following formula: Among them, ht is the output of the LSTM network, S(x, c i ) is the basis spline function, ω i represents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

5. A hydropower generation prediction system based on VMD-LSTM-KAN model, characterized in that: include: Data acquisition module, used to collect historical power generation data and historical meteorological data of the target hydropower station and perform preprocessing; A data processing module is used to perform VMD variational modal decomposition on the pre-processed historical power generation data, and decompose the historical power generation data from time series data into multiple modal components; A feature transformation module, used to combine multiple modal components and historical power generation data into a multi-dimensional output matrix, each dimension in the multi-dimensional output matrix represents a feature of a corresponding modal component or historical power generation data; The data set construction module is used to fuse the pre-processed historical meteorological data with the multi-dimensional output matrix to generate the input data set; The model training module is used to build a prediction model based on the LSTM network, and combine the KAN network to perform nonlinear optimization on the output layer of the LSTM network, and train the built prediction model through the input data set; The prediction module is used to predict the modal component prediction value and power generation prediction value of the target hydropower station in the future period through the trained prediction model.

6. A hydropower generation prediction system based on VMD-LSTM-KAN model according to claim 5, characterized in that: The step of collecting historical power generation data and historical meteorological data of the target hydropower station and preprocessing them includes: Obtain historical power generation data and use the spline interpolation method to fill in the missing data; Obtain historical meteorological data, including relative humidity data, rain water content data, snow water content data and air temperature; The historical power generation data and historical meteorological data are normalized.

7. The hydropower generation prediction system based on the VMD-LSTM-KAN model according to claim 5, characterized in that: In the step of performing VMD variational mode decomposition on the preprocessed historical power generation data: By setting the modal decomposition parameters, the historical power generation data is decomposed into three modal components, corresponding to the low-frequency trend, medium-frequency characteristics and high-frequency noise of the historical power generation data.

8. The hydropower generation prediction system based on VMD-LSTM-KAN model according to claim 5, characterized in that: The method of combining the KAN network with the output layer of the LSTM network for nonlinear optimization is specifically as follows: The output of the LSTM network is received through the KAN network, and the output of the LSTM network is nonlinearly optimized using the following formula: Among them, ht is the output of the LSTM network, S(x, c i ) is the basis spline function, ω i represents the weight parameter of the i-th control point, c i represents the i-th control point, and n is the total number of control points.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the hydropower generation prediction method based on the VMD-LSTM-KAN model as described in any one of claims 1 to 4 is implemented.

Images