A distributed acoustic radar wind power prediction method

Abstract

The present invention relates to the technical field of wind power prediction, and discloses a distributed acoustic radar wind power prediction method, the method comprising: determining a wind speed data sequence decomposition method; constructing an adaptive height wind speed prediction model and a parameter optimization objective function, optimizing the parameters of the adaptive height wind speed prediction model based on the collected wind speed data time series sequence and the parameter optimization objective function, and obtaining an optimal adaptive height wind speed prediction model; inputting the wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model, predicting the wind speed data at different heights, and converting the wind speed data prediction results into wind power prediction results. The present invention obtains wind speed data prediction results by fusing wind speed time series changes and wind speed data spatial information, converts wind speed data prediction results at different heights into wind power prediction results, realizes wind power prediction, and quickly obtains the optimal model by using an adaptive step correction method.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power prediction, and in particular to a distributed acoustic radar wind power prediction method. Background Art

[0002] As one of the most mature and valuable power generation methods in the field of renewable energy, wind power generation has been widely used around the world. With the rapid development of the wind power industry and the continuous increase in annual installed capacity, the proportion of large-scale wind power grid connection is increasing. However, affected by the defects of wind power generation such as randomness, volatility and intermittency, the volatility and randomness in the large-scale grid connection of wind power seriously affect the power system's flow balance, and even affect the safety and reliability of the large power grid during dynamic operation. In more serious cases, a large number of wind farms are disconnected from the grid. In order to effectively solve the problems of "difficulty in grid connection" and "wind abandonment" in the large-scale grid connection of wind power, and to ensure the smooth and reliable grid connection of wind power, wind power prediction has become an important technical means. However, there are large differences in wind power at different heights. The existing method is mainly through interval detection, which has large errors. To address this problem, the present invention proposes a distributed acoustic radar wind power prediction method. Summary of the invention

[0003] In view of this, the present invention provides a distributed acoustic radar wind power prediction method, the purpose of which is: 1) in view of the problem that the fluctuation amplitude of historical wind speed data is large, if the wind speed prediction is performed, there may be a large error, the wind speed data sequences at different heights are sequence decomposed, and different wind speed data sequence components that characterize the wind speed time series change are extracted to reduce the influence of the excessive or too small floating range of the historical wind speed data on the wind speed prediction result, and then an adaptive height wind speed prediction model is constructed, and the wind speed diffusion model is used to perform distributed synthesis on the different wind speed data sequence components to obtain the diffusion space vector of the historical wind speed data in the wind speed data space, and then the wind speed time series change and the wind speed data space information are integrated to obtain the wind speed data prediction result, and the wind speed data prediction results at different heights are converted into wind power prediction results to achieve wind power prediction; 2) the adaptive height wind speed prediction model parameter optimization method with adaptive step correction is used to optimize the parameters, and in the parameter iteration process, the iteration direction is corrected in real time according to the iteration gradient change, the parameter optimization iteration speed is improved, the optimal model parameters are quickly obtained, and the optimal adaptive height wind speed prediction model is constructed, and the parameter iteration step is smoothed by using a smoothing parameter to avoid missing the optimal parameter solution due to excessive iteration step.

[0004] To achieve the above object, the present invention provides a distributed acoustic radar wind power prediction method, comprising the following steps:

[0005] S1: collecting wind speed data time series at different heights to form a wind speed original data series set, wherein the wind speed original data series set includes wind speed data series at different heights;

[0006] S2: performing sequence decomposition on the wind speed data sequences at different heights in the wind speed original data sequence set to obtain wind speed data sequence components;

[0007] S3: Construct an adaptive height wind speed prediction model and a parameter optimization objective function. The constructed model includes two parts: a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and the distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output.

[0008] S4: Optimizing the parameters of the adaptive height wind speed prediction model based on the collected wind speed data time series and the parameter optimization objective function to obtain the optimal adaptive height wind speed prediction model;

[0009] S5: Inputting the current wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model, predicting the wind speed data at different heights, and converting the wind speed data prediction results at different heights into wind power prediction results.

[0010] As a further improvement method of the present invention:

[0011] Optionally, the wind speed data time series sequence collected at different heights in step S1 constitutes a wind speed raw data sequence set, including:

[0012] The wind speed data time series at different heights collected by sodar are used to form a wind speed original data series set, where the wind speed original data series set includes wind speed data series at different heights, and the wind speed original data series set is:

[0013] ;

[0014] in:

[0015] Indicates height The wind speed data series under , H represents the number of categories of collected heights, and the difference between adjacent heights is a preset value , ;

[0016] Indicates at height Down The wind speed data collected at all times, represents the initial time of wind speed data collection, Indicates the cut-off time for wind speed data collection.

[0017] Optionally, in step S2, sequence decomposition of wind speed data sequences at different heights in the wind speed original data sequence set comprises:

[0018] The wind speed data sequences at different heights in the wind speed original data sequence set are decomposed into sequences, wherein the wind speed data sequences in the wind speed original data sequence set are decomposed into sequences. The sequence decomposition process is:

[0019] S21: Traverse the wind speed data sequence , get the peak point of the wind speed data sequence, the peak point includes the maximum peak point and the minimum peak point, where the maximum peak point is defined as:

[0020] like ,but is the maximum peak point, ;

[0021] The minimum peak point is defined as:

[0022] like ,but is the minimum peak point, ;

[0023] S22: Connect all the maximum peak points in the wind speed data sequence to obtain the maximum envelope, connect all the minimum peak points in the wind speed data sequence to obtain the minimum envelope, and obtain the corresponding maximum envelope function based on the envelope And the minimum envelope function , The independent variable of the envelope function corresponds to the time series information of the wind speed data. The solution process of the envelope function based on the envelope line is:

[0024] For the envelope ,in Indicates the first wind speed data in the envelope, n indicates the total number of wind speed data in the envelope, , represents the corresponding time of different wind speed data, and the envelope function of the initialization envelope L :

[0025] ;

[0026] in:

[0027] Represents the independent variable of the envelope function, corresponding to the time series information of the wind speed data. If is less than 0, then ;

[0028] represents the parameters of the envelope function;

[0029] The parameters of the corresponding envelope function are obtained by interpolation, and the envelope based on the wind speed data at n moments is expanded into an envelope function containing wind speed data at several moments. The calculation process of the envelope function based on the interpolation method is as follows:

[0030] ;

[0031] ;

[0032] Solve the above matrix equation to obtain the matrix M, and calculate the envelope function parameters based on the elements in the matrix M, where:

[0033] ;

[0034] ;

[0035] S23: Calculate the mean envelope function: ;

[0036] S24: The mean envelope function As The initial wind speed data sequence component ,in , the wind speed data series Subtract the initial wind speed data sequence component to obtain the residual sequence. If the residual sequence is a non-monotonic sequence, take the residual sequence as the sequence to be traversed, repeat S21-S24, and take the mean envelope function of the residual sequence as the decomposed wind speed data sequence component;

[0037] Until the current residual sequence is a monotonic sequence, the wind speed data sequence is obtained The wind speed data series component vector ,in .

[0038] Optionally, constructing the adaptive height wind speed prediction model in step S3 includes:

[0039] Construct an adaptive height wind speed prediction model, which includes a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and the distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output.

[0040] The wind speed diffusion model is a neural network structure, including three hidden layers, and the wind speed inverse diffusion model is a neural network structure, including two hidden layers and one fully connected layer;

[0041] The wind speed prediction process based on the adaptive height wind speed prediction model is as follows:

[0042] S31: Input all wind speed data sequence components of the wind speed data sequence x at the height and the corresponding height into the wind speed diffusion model, where the wind speed data sequence component vector of the wind speed data sequence x is , The wind speed data sequence x is decomposed into Wind speed data series components;

[0043] S32: Wind speed diffusion model for wind speed data sequence component vector The encoding representation is performed to obtain the diffusion space vector Z, where the calculation formula of the diffusion space vector Z is:

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] in:

[0049] represents the diffusion space component corresponding to the k-th wind speed data sequence component, ,correspond Distribution satisfy ,in represents the identity matrix;

[0050] represents the added noise, Indicates noise It conforms to the Gaussian distribution. Represents element-wise multiplication;

[0051] represents the weight coefficients of the three hidden layers in the wind speed diffusion model, Represents the bias of the three hidden layers in the wind speed diffusion model;

[0052] S33: inputting the diffusion space vector Z into the wind speed inverse diffusion model, the wind speed inverse diffusion model calculates the wind speed data corresponding to the diffusion space vector Z as the prediction result, wherein the calculation formula of the wind speed inverse diffusion model is:

[0053] ;

[0054] in:

[0055] represents the weight coefficient of the two hidden layers in the wind speed inverse diffusion model, Represents the bias of the two hidden layers in the wind speed inverse diffusion model;

[0056] Y represents the wind speed data corresponding to the diffusion space vector Z, that is, the predicted wind speed data;

[0057] represents the activation function;

[0058] W represents the weight of the fully connected layer in the wind speed inverse diffusion model, and T represents transpose.

[0059] Optionally, the parameter optimization objective function of the adaptive height wind speed prediction model is constructed in step S3, including:

[0060] Construct the parameter optimization objective function of the adaptive height wind speed prediction model, where the parameters to be optimized are , represents the parameters to be optimized in the wind speed diffusion model, represents the parameters to be optimized of the wind speed inverse diffusion model, and T represents transposition; the wind speed data sequences collected at different heights are processed to obtain the training set of the height wind speed prediction model, where the processing flow is:

[0061] The wind speed data at the last moment in the historical wind speed data sequence collected at different heights is taken as the true wind speed data, and the remaining wind speed data is taken as the historical wind speed data sequence. The historical wind speed data sequence is sequence decomposed to obtain the wind speed data sequence component vector corresponding to the historical wind speed data sequence. The wind speed data sequence component vector and the true wind speed data constitute a set of training data, and the processed historical wind speed data sequence is reprocessed until U sets of training data are obtained at each height. The obtained U The H group of training data constitutes the training set of the height wind speed prediction model;

[0062] The constructed parameter optimization objective function is:

[0063] ;

[0064] in:

[0065] Indicates that Reach the minimum parameters ;

[0066] Indicates height The actual wind speed data of the u-th group of training data;

[0067] Indicates that the height The wind speed data sequence component vector in the u-th group of training data is input to the parameter-based In the adaptive height wind speed prediction model, the model outputs the predicted wind speed data.

[0068] Optionally, in step S4, the parameters of the adaptive height wind speed prediction model are optimized based on the collected wind speed data time series and the parameter optimization objective function, including:

[0069] Based on the collected wind speed data time series and the parameter optimization objective function, the parameters of the adaptive height wind speed prediction model are optimized, and the optimal adaptive height wind speed prediction model is constructed based on the optimized parameters. The parameter optimization process of the adaptive height wind speed prediction model is as follows:

[0070] S41: Generate a set of adaptive height wind speed prediction model parameters ;

[0071] S42: Set the current iteration number of parameter optimization to d, the initial value of d is 0, and the maximum iteration number is Max. Then the adaptive height wind speed prediction model parameters of the dth iteration are ;

[0072] S43: Substitute into the parameter optimization objective function and calculate The corresponding parameter gradient ,like Less than the preset threshold Or the maximum number of iterations is reached, the iteration is terminated and the As the optimized parameters, represents the L1 norm;

[0073] S44: Calculate the parameters The iteration step length :

[0074] ;

[0075] ;

[0076] ;

[0077] in:

[0078] Iteration step length Direction control parameters, Represents the iteration step size based on iteration direction control Range control parameters;

[0079] represents the smoothing index, set it to 0.9;

[0080] S45: Parameter-based The iteration step length , for the parameters To iterate:

[0081] ;

[0082] make , return to step S43.

[0083] Optionally, in step S5, the wind speed data sequence components at different current heights are input into the optimal adaptive height wind speed prediction model to predict the wind speed data at different heights, including:

[0084] Collect the current wind speed data time series historical data at different heights to form the corresponding wind speed data sequence, and decompose the wind speed data sequence at different heights to obtain the wind speed data sequence components at different heights. Input the wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model in turn to obtain the wind speed data prediction results at different heights:

[0085] ;

[0086] in:

[0087] Indicates height The wind speed data prediction results are as follows.

[0088] Optionally, in step S5, converting the wind speed data prediction results at different heights into wind power prediction results includes:

[0089] Collect the heights of the centers of different wind turbines from the ground, where the height of the center of any j-th wind turbine from the ground is , , represents the collected height range set, then the wind power prediction result of any j-th wind turbine is:

[0090] ;

[0091] in:

[0092] Indicates the blade area of ​​the wind turbine;

[0093] Indicates the air density;

[0094] Represents the wind power prediction result of the j-th wind turbine.

[0095] In an embodiment of the present invention, sonar is used to obtain historical wind speed data, and wind speed data at different heights are predicted based on the historical wind speed data. Wind turbines at different heights are matched with the predicted results of wind speed data at corresponding heights to obtain wind power prediction results of the wind turbines, thereby realizing wind power prediction based on sonar.

[0096] In order to solve the above problem, the present invention provides an electronic device, the electronic device comprising:

[0097] A memory storing at least one instruction;

[0098] Communication interface, enabling electronic equipment to communicate; and

[0099] The processor executes the instructions stored in the memory to implement the above-mentioned distributed acoustic radar wind power prediction method.

[0100] In order to solve the above problem, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned distributed acoustic radar wind power prediction method.

[0101] Compared with the prior art, the present invention proposes a distributed acoustic radar wind power prediction method, which has the following advantages:

[0102] First, this scheme proposes a wind speed data sequence decomposition method, which performs sequence decomposition on the wind speed data sequences at different heights in the wind speed original data sequence set, wherein the wind speed data sequences in the wind speed original data sequence set are The sequence decomposition process is: traverse the wind speed data sequence , get the peak point of the wind speed data sequence, the peak point includes the maximum peak point and the minimum peak point; connect all the maximum peak points in the wind speed data sequence to get the maximum envelope, connect all the minimum peak points in the wind speed data sequence to get the minimum envelope, and get the corresponding maximum envelope function based on the envelope And the minimum envelope function , The independent variable representing the envelope function corresponds to the time series information of the wind speed data, and the mean envelope function is calculated:

[0103] ; The mean envelope function As The initial wind speed data sequence component ,in , the wind speed data series Subtract the initial wind speed data sequence component to get the residual sequence. If the residual sequence is a non-monotonic sequence, take the residual sequence as the sequence to be traversed and repeat the sequence decomposition until the current residual sequence is a monotonic sequence. Take the mean envelope function of the residual sequence as the decomposed wind speed data sequence component to get the wind speed data sequence The wind speed data series component vector ,in In view of the large fluctuation range of historical wind speed data, if wind speed prediction is performed, there may be large errors. This scheme decomposes the wind speed data sequences at different heights, extracts different wind speed data sequence components that characterize the time series changes of wind speed, and reduces the impact of the excessive or small floating range of historical wind speed data on the wind speed prediction results. Then, an adaptive height wind speed prediction model is constructed, and the wind speed diffusion model is used to perform distributed synthesis of different wind speed data sequence components to obtain the diffusion space vector of the historical wind speed data in the wind speed data space. Then, the wind speed time series changes and the wind speed data spatial information are integrated to obtain the wind speed data prediction results, and the wind speed data prediction results at different heights are converted into wind power prediction results to achieve wind power prediction.

[0104] At the same time, this scheme proposes a parameter optimization method for an adaptive height wind speed prediction model with adaptive step size correction. The constructed parameter optimization objective function is:

[0105] ;

[0106] in: Indicates that Reach the minimum parameters ; Indicates height The actual wind speed data of the u-th group of training data; Indicates that the height The wind speed data sequence component vector in the u-th group of training data is input to the parameter-based In the adaptive height wind speed prediction model, the model outputs the predicted wind speed data. Based on the collected wind speed data time series and the parameter optimization objective function, the parameters of the adaptive height wind speed prediction model are optimized, and the optimal adaptive height wind speed prediction model is constructed based on the optimized parameters. The parameter optimization process of the adaptive height wind speed prediction model is as follows: Generate a set of adaptive height wind speed prediction model parameters ; Set the current iteration number of parameter optimization to d, the initial value of d is 0, and the maximum iteration number is Max. Then the parameters of the adaptive height wind speed prediction model for the dth iteration are ;Will Substitute into the parameter optimization objective function and calculate The corresponding parameter gradient ,like Less than the preset threshold Or the maximum number of iterations is reached, the iteration is terminated and the As the optimized parameters, Represents the L1 norm; the parameters are calculated The iteration step length :

[0107] ;

[0108] ;

[0109] ;

[0110] in: Iteration step length Direction control parameters, Represents the iteration step size based on iteration direction control Range control parameters; represents the smoothing index, which is set to 0.9; based on the parameter The iteration step length , for the parameters To iterate:

[0111] ;

[0112] make Perform parameter iteration. This scheme uses the adaptive height wind speed prediction model parameter optimization method with adaptive step correction to optimize parameters. During the parameter iteration process, the iteration direction is corrected in real time according to the iterative gradient change to improve the parameter optimization iteration speed, quickly obtain the optimal model parameters and build the optimal adaptive height wind speed prediction model. The smoothing parameter is used to make the parameter iteration step smooth to avoid missing the optimal parameter solution due to too large an iteration step. BRIEF DESCRIPTION OF THE DRAWINGS

[0113] Figure 1 A schematic diagram of a flow chart of a distributed acoustic radar wind power prediction method provided by an embodiment of the present invention;

[0114] Figure 2 A schematic diagram of the structure of an electronic device for implementing a distributed acoustic radar wind power prediction method provided by an embodiment of the present invention.

[0115] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. Implementation

[0116] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0117] The embodiment of the present application provides a distributed acoustic radar wind power prediction method. The execution subject of the distributed acoustic radar wind power prediction method includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the distributed acoustic radar wind power prediction method can be executed by software or hardware installed in a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. Example

[0118] S1: Collect wind speed data time series at different heights to form a wind speed original data series set, wherein the wind speed original data series set includes wind speed data series at different heights.

[0119] The wind speed data time series sequence collected at different heights in step S1 constitutes a wind speed raw data sequence set, including:

[0120] The wind speed data time series at different heights collected by sodar are used to form a wind speed original data series set, where the wind speed original data series set includes wind speed data series at different heights, and the wind speed original data series set is:

[0121] ;

[0122] in:

[0123] Indicates height The wind speed data series under , H represents the number of categories of collected heights, and the difference between adjacent heights is a preset value , ;

[0124] Indicates at height Down The wind speed data collected at all times, represents the initial time of wind speed data collection, Indicates the cut-off time for wind speed data collection.

[0125] S2: performing sequence decomposition on wind speed data sequences at different heights in the wind speed original data sequence set to obtain wind speed data sequence components.

[0126] In the step S2, the wind speed data sequences at different heights in the wind speed original data sequence set are subjected to sequence decomposition, including:

[0127] The wind speed data sequences at different heights in the wind speed original data sequence set are decomposed into sequences, wherein the wind speed data sequences in the wind speed original data sequence set are decomposed into sequences. The sequence decomposition process is:

[0128] S21: Traverse the wind speed data sequence , get the peak point of the wind speed data sequence, the peak point includes the maximum peak point and the minimum peak point, where the maximum peak point is defined as:

[0129] like ,but is the maximum peak point, ;

[0130] The minimum peak point is defined as:

[0131] like ,but is the minimum peak point, ;

[0132] S22: Connect all the maximum peak points in the wind speed data sequence to obtain the maximum envelope, connect all the minimum peak points in the wind speed data sequence to obtain the minimum envelope, and obtain the corresponding maximum envelope function based on the envelope And the minimum envelope function , The independent variable of the envelope function corresponds to the time series information of the wind speed data. The solution process of the envelope function based on the envelope line is:

[0133] For the envelope ,in Indicates the first wind speed data in the envelope, n indicates the total number of wind speed data in the envelope, , represents the corresponding time of different wind speed data, and the envelope function of the initialization envelope L :

[0134] ;

[0135] in:

[0136] Represents the independent variable of the envelope function, corresponding to the time series information of the wind speed data. If is less than 0, then ;

[0137] represents the parameters of the envelope function;

[0138] The parameters of the corresponding envelope function are obtained by interpolation, and the envelope based on the wind speed data at n moments is expanded into an envelope function containing wind speed data at several moments. The calculation process of the envelope function based on the interpolation method is as follows:

[0139] ;

[0140] ;

[0141] Solve the above matrix equation to obtain the matrix M, and calculate the envelope function parameters based on the elements in the matrix M, where:

[0142] ;

[0143] ;

[0144] S23: Calculate the mean envelope function: ;

[0145] S24: The mean envelope function As The initial wind speed data sequence component ,in , the wind speed data series Subtract the initial wind speed data sequence component to obtain the residual sequence. If the residual sequence is a non-monotonic sequence, take the residual sequence as the sequence to be traversed, repeat S21-S24, and take the mean envelope function of the residual sequence as the decomposed wind speed data sequence component;

[0146] Until the current residual sequence is a monotonic sequence, the wind speed data sequence is obtained The wind speed data series component vector ,in .

[0147] S3: Construct an adaptive height wind speed prediction model and a parameter optimization objective function. The constructed model consists of two parts: a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and a distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output.

[0148] The step S3 constructs an adaptive height wind speed prediction model, including:

[0149] Construct an adaptive height wind speed prediction model, which includes a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and the distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output.

[0150] The wind speed diffusion model is a neural network structure, including three hidden layers, and the wind speed inverse diffusion model is a neural network structure, including two hidden layers and one fully connected layer;

[0151] The wind speed prediction process based on the adaptive height wind speed prediction model is as follows:

[0152] S31: Input all wind speed data sequence components of the wind speed data sequence x at the height and the corresponding height into the wind speed diffusion model, where the wind speed data sequence component vector of the wind speed data sequence x is , The wind speed data sequence x is decomposed into Wind speed data series components;

[0153] S32: Wind speed diffusion model for wind speed data sequence component vector The encoding representation is performed to obtain the diffusion space vector Z, where the calculation formula of the diffusion space vector Z is:

[0154] ;

[0155] ;

[0156] ;

[0157] ;

[0158] in:

[0159] represents the diffusion space component corresponding to the k-th wind speed data sequence component, ,correspond Distribution satisfy ,in represents the identity matrix;

[0160] represents the added noise, Indicates noise It conforms to the Gaussian distribution. Represents element-wise multiplication;

[0161] represents the weight coefficients of the three hidden layers in the wind speed diffusion model, Represents the bias of the three hidden layers in the wind speed diffusion model;

[0162] S33: inputting the diffusion space vector Z into the wind speed inverse diffusion model, the wind speed inverse diffusion model calculates the wind speed data corresponding to the diffusion space vector Z as the prediction result, wherein the calculation formula of the wind speed inverse diffusion model is:

[0163] ;

[0164] represents the weight coefficient of the two hidden layers in the wind speed inverse diffusion model, Represents the bias of the two hidden layers in the wind speed inverse diffusion model;

[0165] Y represents the wind speed data corresponding to the diffusion space vector Z, that is, the predicted wind speed data;

[0166] represents the activation function;

[0167] W represents the weight of the fully connected layer in the wind speed inverse diffusion model, and T represents transpose.

[0168] The parameter optimization objective function of the adaptive height wind speed prediction model is constructed in step S3, including:

[0169] Construct the parameter optimization objective function of the adaptive height wind speed prediction model, where the parameters to be optimized are , represents the parameters to be optimized in the wind speed diffusion model, represents the parameters to be optimized in the wind speed inverse diffusion model, and T represents transposition;

[0170] The collected wind speed data sequences at different heights are processed to obtain the training set of the height wind speed prediction model. The processing flow is as follows:

[0171] The wind speed data at the last moment in the historical wind speed data sequence collected at different heights is taken as the true wind speed data, and the remaining wind speed data is taken as the historical wind speed data sequence. The historical wind speed data sequence is sequence decomposed to obtain the wind speed data sequence component vector corresponding to the historical wind speed data sequence. The wind speed data sequence component vector and the true wind speed data constitute a set of training data, and the processed historical wind speed data sequence is reprocessed until U sets of training data are obtained at each height. The obtained U The H group of training data constitutes the training set of the height wind speed prediction model;

[0172] The constructed parameter optimization objective function is:

[0173] ;

[0174] in:

[0175] Indicates that Reach the minimum parameters ;

[0176] Indicates height The actual wind speed data of the u-th group of training data;

[0177] Indicates that the height The wind speed data sequence component vector in the u-th group of training data is input to the parameter-based In the adaptive height wind speed prediction model, the model outputs the predicted wind speed data.

[0178] S4: Optimize the parameters of the adaptive height wind speed prediction model based on the collected wind speed data time series and the parameter optimization objective function to obtain the optimal adaptive height wind speed prediction model.

[0179] In the step S4, the parameters of the adaptive height wind speed prediction model are optimized based on the collected wind speed data time series and the parameter optimization objective function, including:

[0180] Based on the collected wind speed data time series and the parameter optimization objective function, the parameters of the adaptive height wind speed prediction model are optimized, and the optimal adaptive height wind speed prediction model is constructed based on the optimized parameters. The parameter optimization process of the adaptive height wind speed prediction model is as follows:

[0181] S41: Generate a set of adaptive height wind speed prediction model parameters ;

[0182] S42: Set the current iteration number of parameter optimization to d, the initial value of d is 0, and the maximum iteration number is Max. Then the adaptive height wind speed prediction model parameters of the dth iteration are ;

[0183] S43: Substitute into the parameter optimization objective function and calculate The corresponding parameter gradient ,like Less than the preset threshold Or the maximum number of iterations is reached, the iteration is terminated and the As the optimized parameters, represents the L1 norm;

[0184] S44: Calculate the parameters The iteration step length :

[0185] ;

[0186] ;

[0187] ;

[0188] in:

[0189] Iteration step length Direction control parameters, Represents the iteration step size based on iteration direction control Range control parameters;

[0190] represents the smoothing index, set it to 0.9;

[0191] S45: Parameter-based The iteration step length , for the parameters To iterate:

[0192] ;

[0193] make , return to step S43.

[0194] S5: Inputting the current wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model, predicting the wind speed data at different heights, and converting the wind speed data prediction results at different heights into wind power prediction results.

[0195] In the step S5, the wind speed data sequence components at different heights are input into the optimal adaptive height wind speed prediction model to predict the wind speed data at different heights, including:

[0196] Collect the current wind speed data time series historical data at different heights to form the corresponding wind speed data sequence, and decompose the wind speed data sequence at different heights to obtain the wind speed data sequence components at different heights. Input the wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model in turn to obtain the wind speed data prediction results at different heights:

[0197] ;

[0198] in:

[0199] Indicates height The wind speed data prediction results are as follows.

[0200] The step S5 converts the wind speed data prediction results at different heights into wind power prediction results, including:

[0201] Collect the heights of the centers of different wind turbines from the ground, where the height of the center of any j-th wind turbine from the ground is , , represents the collected height range set, then the wind power prediction result of any j-th wind turbine is:

[0202] ;

[0203] in:

[0204] Indicates the blade area of ​​the wind turbine;

[0205] Indicates the air density;

[0206] Represents the wind power prediction result of the j-th wind turbine.

[0207] Embodiment 2:

[0208] like Figure 2 FIG. 1 is a schematic diagram of the structure of an electronic device for implementing a distributed acoustic radar wind power prediction method provided by an embodiment of the present invention.

[0209] The electronic device 1 may include a processor 10 , a memory 11 , a communication interface 13 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10 , such as a program 12 .

[0210] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 may not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the program 12, etc., but also be used to temporarily store data that has been output or is to be output.

[0211] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules (such as the program 12 for implementing distributed acoustic radar wind power prediction) stored in the memory 11, and calls data stored in the memory 11, so as to execute various functions of the electronic device 1 and process data.

[0212] The communication interface 13 may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices, and to achieve connection and communication between internal components of the electronic device.

[0213] The bus may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0214] Figure 2 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 2 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0215] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0216] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0217] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0218] The program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0219] The wind speed data time series at different heights are collected to form a wind speed original data series set, wherein the wind speed original data series set includes wind speed data series at different heights;

[0220] Decomposing the wind speed data sequences at different heights in the wind speed original data sequence set to obtain wind speed data sequence components;

[0221] Construct an adaptive height wind speed prediction model and parameter optimization objective function;

[0222] Based on the collected wind speed data time series and the parameter optimization objective function, the parameters of the adaptive height wind speed prediction model are optimized to obtain the optimal adaptive height wind speed prediction model;

[0223] The wind speed data sequence components at different current heights are input into the optimal adaptive height wind speed prediction model to predict the wind speed data at different heights, and the wind speed data prediction results at different heights are converted into wind power prediction results.

[0224] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figure 1 to Figure 2 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0225] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0226] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0227] The above are only preferred 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 distributed acoustic radar wind power prediction method, characterized in that: The method comprises: S1: collecting wind speed data time series at different heights to form a wind speed original data series set, wherein the wind speed original data series set includes wind speed data series at different heights; S2: performing sequence decomposition on the wind speed data sequences at different heights in the wind speed original data sequence set to obtain wind speed data sequence components; S3: Construct an adaptive height wind speed prediction model and a parameter optimization objective function. The constructed model includes two parts: a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and the distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output. S4: Optimizing the parameters of the adaptive height wind speed prediction model based on the collected wind speed data time series and the parameter optimization objective function to obtain the optimal adaptive height wind speed prediction model; S5: Inputting the current wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model, predicting the wind speed data at different heights, and converting the wind speed data prediction results at different heights into wind power prediction results.

2. A distributed acoustic radar wind power prediction method as claimed in claim 1, characterized in that: The wind speed data time series sequence collected at different heights in step S1 constitutes a wind speed raw data sequence set, including: The wind speed data time series at different heights collected by sodar are used to form a wind speed original data series set, where the wind speed original data series set includes wind speed data series at different heights, and the wind speed original data series set is: ; in: Indicates height The wind speed data series under , H represents the number of categories of collected heights, and the difference between adjacent heights is a preset value , ; Indicates at height Down The wind speed data collected at all times, represents the initial time of wind speed data collection, Indicates the cut-off time for wind speed data collection.

3. A distributed acoustic radar wind power prediction method as claimed in claim 2, characterized in that: In the step S2, the wind speed data sequences at different heights in the wind speed original data sequence set are subjected to sequence decomposition, including: The wind speed data sequences at different heights in the wind speed original data sequence set are decomposed into sequences, wherein the wind speed data sequences in the wind speed original data sequence set are decomposed into sequences. The sequence decomposition process is: S21: Traverse the wind speed data sequence , get the peak point of the wind speed data sequence, the peak point includes the maximum peak point and the minimum peak point, where the maximum peak point is defined as: like ,but is the maximum peak point, ; The minimum peak point is defined as: like ,but is the minimum peak point, ; S22: Connect all the maximum peak points in the wind speed data sequence to obtain the maximum envelope, connect all the minimum peak points in the wind speed data sequence to obtain the minimum envelope, and obtain the corresponding maximum envelope function based on the envelope And the minimum envelope function , The independent variable of the envelope function corresponds to the time series information of the wind speed data. The solution process of the envelope function based on the envelope line is: For the envelope ,in Indicates the first wind speed data in the envelope, n indicates the total number of wind speed data in the envelope, , represents the corresponding time of different wind speed data, and the envelope function of the initialization envelope L : ; in: Represents the independent variable of the envelope function, corresponding to the time series information of the wind speed data. If is less than 0, then ; represents the parameters of the envelope function; The parameters of the corresponding envelope function are obtained by interpolation, and the envelope based on the wind speed data at n moments is expanded into an envelope function containing wind speed data at several moments. The calculation process of the envelope function based on the interpolation method is as follows: ; ; Solve the above matrix equation to obtain the matrix M, and calculate the envelope function parameters based on the elements in the matrix M, where: ; ; S23: Calculate the mean envelope function: ; S24: The mean envelope function As The initial wind speed data sequence component ,in , the wind speed data series Subtract the initial wind speed data sequence component to obtain the residual sequence. If the residual sequence is a non-monotonic sequence, take the residual sequence as the sequence to be traversed, repeat S21-S24, and take the mean envelope function of the residual sequence as the decomposed wind speed data sequence component; Until the current residual sequence is a monotonic sequence, the wind speed data sequence is obtained The wind speed data series component vector ,in .

4. A distributed acoustic radar wind power prediction method as claimed in claim 1, characterized in that: The step S3 constructs an adaptive height wind speed prediction model, including: Construct an adaptive height wind speed prediction model, which includes a wind speed diffusion model and a wind speed inverse diffusion model. The wind speed diffusion model takes the height and all wind speed data sequence components at the corresponding height as input and the distributed synthetic diffusion space vector as output. The wind speed inverse diffusion model takes the diffusion space vector as input and the predicted wind speed data as output. The wind speed diffusion model is a neural network structure, including three hidden layers, and the wind speed inverse diffusion model is a neural network structure, including two hidden layers and one fully connected layer; The wind speed prediction process based on the adaptive height wind speed prediction model is as follows: S31: Input all wind speed data sequence components of the wind speed data sequence x at the height and the corresponding height into the wind speed diffusion model, where the wind speed data sequence component vector of the wind speed data sequence x is , The wind speed data sequence x is decomposed into Wind speed data series components; S32: Wind speed diffusion model for wind speed data sequence component vector The encoding representation is performed to obtain the diffusion space vector Z, where the calculation formula of the diffusion space vector Z is: ; ; ; ; in: represents the diffusion space component corresponding to the k-th wind speed data sequence component, ,correspond Distribution satisfy ,in represents the identity matrix; represents the added noise, Indicates noise It conforms to the Gaussian distribution. Represents element-wise multiplication; represents the weight coefficients of the three hidden layers in the wind speed diffusion model, Represents the bias of the three hidden layers in the wind speed diffusion model; S33: inputting the diffusion space vector Z into the wind speed inverse diffusion model, the wind speed inverse diffusion model calculates the wind speed data corresponding to the diffusion space vector Z as the prediction result, wherein the calculation formula of the wind speed inverse diffusion model is: ; in: represents the weight coefficient of the two hidden layers in the wind speed inverse diffusion model, Represents the bias of the two hidden layers in the wind speed inverse diffusion model; Y represents the wind speed data corresponding to the diffusion space vector Z, that is, the predicted wind speed data; represents the activation function; W represents the weight of the fully connected layer in the wind speed inverse diffusion model, and T represents transpose.

5. A distributed acoustic radar wind power prediction method as claimed in claim 4, characterized in that: The parameter optimization objective function of the adaptive height wind speed prediction model is constructed in step S3, including: Construct the parameter optimization objective function of the adaptive height wind speed prediction model, where the parameters to be optimized are , represents the parameters to be optimized in the wind speed diffusion model, represents the parameters to be optimized in the wind speed inverse diffusion model, and T represents transposition; The collected wind speed data sequences at different heights are processed to obtain the training set of the height wind speed prediction model. The processing flow is as follows: The wind speed data at the last moment in the historical wind speed data sequence collected at different heights is taken as the true wind speed data, and the remaining wind speed data is taken as the historical wind speed data sequence. The historical wind speed data sequence is sequence decomposed to obtain the wind speed data sequence component vector corresponding to the historical wind speed data sequence. The wind speed data sequence component vector and the true wind speed data constitute a set of training data, and the processed historical wind speed data sequence is reprocessed until U sets of training data are obtained at each height. The obtained U The H group of training data constitutes the training set of the height wind speed prediction model; The constructed parameter optimization objective function is: ; in: Indicates that Reach the minimum parameters ; Indicates height The actual wind speed data of the u-th group of training data; Indicates that the height The wind speed data sequence component vector in the u-th group of training data is input to the parameter-based In the adaptive height wind speed prediction model, the model outputs the predicted wind speed data.

6. A distributed acoustic radar wind power prediction method as claimed in claim 5, characterized in that: In the step S4, the parameters of the adaptive height wind speed prediction model are optimized based on the collected wind speed data time series and the parameter optimization objective function, including: The parameters of the adaptive height wind speed prediction model are optimized based on the collected wind speed data time series and the parameter optimization objective function, and the optimal adaptive height wind speed prediction model is constructed based on the optimized parameters. The parameter optimization process of the adaptive height wind speed prediction model is as follows: S41: Generate a set of adaptive height wind speed prediction model parameters ; S42: Set the current iteration number of parameter optimization to d, the initial value of d is 0, and the maximum iteration number is Max. Then the adaptive height wind speed prediction model parameters of the dth iteration are ; S43: Substitute into the parameter optimization objective function and calculate The corresponding parameter gradient ,like Less than the preset threshold Or the maximum number of iterations is reached, the iteration is terminated and the As the optimized parameters, represents the L1 norm; S44: Calculate the parameters The iteration step length : ; ; ; in: Iteration step length Direction control parameters, Represents the iteration step size based on iteration direction control Range control parameters; represents the smoothing index, set it to 0.9; S45: Parameter-based The iteration step length , for the parameters To iterate: ; make , return to step S43.

7. A distributed acoustic radar wind power prediction method as claimed in claim 6, characterized in that: In the step S5, the wind speed data sequence components at different heights are input into the optimal adaptive height wind speed prediction model to predict the wind speed data at different heights, including: Collect the current wind speed data time series historical data at different heights to form the corresponding wind speed data sequence, and decompose the wind speed data sequence at different heights to obtain the wind speed data sequence components at different heights. Input the wind speed data sequence components at different heights into the optimal adaptive height wind speed prediction model in turn to obtain the wind speed data prediction results at different heights: ; in: Indicates height The wind speed data prediction results are as follows.

8. A distributed acoustic radar wind power prediction method as claimed in claim 7, characterized in that: The step S5 converts the wind speed data prediction results at different heights into wind power prediction results, including: Collect the heights of the centers of different wind turbines from the ground, where the height of the center of any j-th wind turbine from the ground is , , represents the collected height range set, then the wind power prediction result of any j-th wind turbine is: ; in: Indicates the blade area of ​​the wind turbine; Indicates the air density; Represents the wind power prediction result of the j-th wind turbine.

Images