Wind energy prediction method, system, device and medium based on multi-source data fusion

Abstract

The application provides a wind energy prediction method, system, device and medium based on multi-source data fusion, and belongs to the technical field of wind power. The method comprises the following steps: acquiring a plurality of meteorological data sets corresponding to a to-be-tested wind farm in a historical time period provided by a plurality of data sources; preprocessing the plurality of meteorological data sets to obtain a plurality of to-be-tested meteorological data sets; inputting the plurality of to-be-tested meteorological data sets into a wind energy prediction model, wherein the wind energy prediction model comprises an independent prediction module and a fusion prediction module; processing the plurality of to-be-tested meteorological data sets by using the independent prediction module to obtain a plurality of initial prediction wind energy data sets corresponding to the to-be-tested wind farm in a future time period; and aligning, splicing and analyzing the plurality of initial prediction wind energy data sets by using the fusion prediction module to obtain a final prediction wind energy data set of the to-be-tested wind farm in the future time period. The application can realize more accurate prediction of wind energy data of the to-be-tested wind farm in the future time period.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power, in particular to a wind energy prediction method, system, device and medium based on multi-source data fusion. BACKGROUND

[0002] In the prior art, a machine learning model is usually used to analyze the wind farm related historical data provided by a single data source, so as to predict the future wind energy data of the wind farm. However, this implementation has certain limitations. If the wind farm related historical data provided by the single data source is not accurate and comprehensive enough, the credibility of the future wind energy data of the wind farm predicted by the model is not high. SUMMARY

[0003] The main purpose of the present application is to provide a wind energy prediction method, system, device and medium based on multi-source data fusion. By using a specific wind energy prediction model to analyze and fuse the related meteorological data of a to-be-tested wind farm provided by multiple data sources in a historical time period, the wind energy data of the to-be-tested wind farm in a future time period can be accurately predicted.

[0004] To achieve the above purpose, one aspect of the present application provides a wind energy prediction method based on multi-source data fusion, which comprises:

[0005] Obtaining a plurality of meteorological data sets corresponding to a to-be-tested wind farm in a historical time period provided by a plurality of data sources;

[0006] Preprocessing the plurality of meteorological data sets to obtain a plurality of to-be-tested meteorological data sets corresponding thereto;

[0007] Inputting the plurality of to-be-tested meteorological data sets into a wind energy prediction model, the wind energy prediction model comprising an independent prediction module and a fusion prediction module, using the independent prediction module to process the plurality of to-be-tested meteorological data sets respectively to obtain a plurality of initial predicted wind energy data sets corresponding to the to-be-tested wind farm in a future time period, and using the fusion prediction module to align, splice and analyze the plurality of initial predicted wind energy data sets to obtain a final predicted wind energy data set of the to-be-tested wind farm in the future time period.

[0008] Further, the preprocessing of the plurality of meteorological data sets to obtain a plurality of to-be-tested meteorological data sets corresponding thereto comprises:

[0009] Normalizing the plurality of meteorological data sets to obtain a plurality of initial meteorological data sets corresponding thereto;

[0010] For each of the initial meteorological data sets, the initial meteorological data sets are divided according to a preset time interval to obtain a plurality of initial meteorological data subsets to form a corresponding to-be-measured meteorological data set; wherein the preset time interval is greater than or equal to the length of the future time period.

[0011] Further, the independent prediction module includes a plurality of independent prediction sub-modules corresponding to the plurality of data sources, and the network structures of each of the independent prediction sub-modules are the same.

[0012] For the independent prediction sub-module set for each of the data sources, the independent prediction sub-module includes a spatial attention unit, a convolution unit and a long short-term memory network unit; when the independent prediction sub-module receives the to-be-measured meteorological data set associated with the data source, the spatial attention unit is used to perform preliminary feature extraction on the to-be-measured meteorological data set to obtain key feature data; the convolution unit is used to perform re-feature extraction on the key feature data to obtain first deep feature data; and the long short-term memory network unit is used to analyze the first deep feature data to obtain the initial prediction wind energy data set of the to-be-measured wind farm in the future time period.

[0013] Further, the fusion prediction module includes an alignment layer, a splicing layer, a convolution layer and a multi-layer perception machine.

[0014] When the fusion prediction module receives the plurality of initial prediction wind energy data sets, the alignment layer is used to respectively perform time interpolation on the plurality of initial prediction wind energy data sets according to the length of the future time period to obtain a plurality of first prediction wind energy data sets; the splicing layer is used to splice the plurality of first prediction wind energy data sets to obtain a second prediction wind energy data set after fusion; the convolution layer is used to extract features from the second prediction wind energy data set to obtain second deep feature data; and the multi-layer perception machine is used to analyze the second deep feature data to obtain the final prediction wind energy data set of the to-be-measured wind farm in the future time period.

[0015] Further, the wind energy prediction model is trained in the following way:

[0016] A plurality of wind energy data sets corresponding to a plurality of target wind farms in a first time period are obtained.

[0017] The plurality of wind energy data sets are preprocessed to obtain a plurality of first wind energy data sets.

[0018] obtaining a plurality of first meteorological data sets corresponding to the plurality of target wind farms provided by the plurality of data sources in a second time period; wherein the second time period is before the first time period, the length of the second time period is the same as the length of the historical time period, and the length of the first time period is the same as the length of the future time period;

[0019] preprocessing the plurality of first meteorological data sets to obtain a plurality of second meteorological data sets;

[0020] combining the plurality of first wind energy data sets as training labels with the plurality of second meteorological data sets to train an initial wind energy prediction model pre-built to obtain the wind energy prediction model.

[0021] Further, the preprocessing of the plurality of wind energy data sets to obtain a plurality of first wind energy data sets comprises:

[0022] stationary processing the plurality of wind energy data sets to obtain a plurality of initial wind energy data sets;

[0023] For each of the initial wind energy data sets, the initial wind energy data set contains wind power data and wind speed data, and according to the on-line capacity of the target wind farm associated with the initial wind energy data set, the wind power data is normalized to obtain first wind power data;

[0024] combining the wind speed data and the first wind power data to form a corresponding first wind energy data set.

[0025] Further, the preprocessing of the plurality of wind energy data sets to obtain a plurality of first wind energy data sets comprises:

[0026] normalizing the plurality of first meteorological data sets to obtain a plurality of first initial meteorological data sets;

[0027] For each of the first initial meteorological data sets, the first initial meteorological data set is divided according to the preset time interval to obtain a plurality of first initial meteorological data subsets to form a corresponding second meteorological data set.

[0028] To achieve the above purpose, another aspect of the present application provides a wind energy prediction system based on multi-source data fusion, the system comprises:

[0029] an acquisition module for obtaining a plurality of meteorological data sets corresponding to a target wind farm in a historical time period provided by a plurality of data sources;

[0030] a preprocessing module for preprocessing the plurality of meteorological data sets to obtain a plurality of target meteorological data sets;

[0031] The analysis module is configured to input the plurality of to-be-tested meteorological data sets into a wind energy prediction model, the wind energy prediction model comprising an independent prediction module and a fusion prediction module, the independent prediction module being configured to process the plurality of to-be-tested meteorological data sets respectively to obtain a plurality of initial prediction wind energy data sets corresponding to the to-be-tested wind farm in a future time period, and the fusion prediction module being configured to align and splice and analyze the plurality of initial prediction wind energy data sets to obtain a final prediction wind energy data set of the to-be-tested wind farm in the future time period.

[0032] To achieve the above object, another aspect of the present application provides an electronic device, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above method when executing the computer program.

[0033] To achieve the above object, another aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the above method when executed by a processor.

[0034] The present application at least has the following beneficial effects: by setting an independent prediction module and a fusion prediction module in a specific wind energy prediction model, the related meteorological data of a to-be-tested wind farm in a historical time period provided by a plurality of data sources can be independently predicted and fusion processed, and then the wind energy data of the to-be-tested wind farm in a future time period can be more accurately predicted. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a flowchart of a wind energy prediction method based on multi-source data fusion provided by an embodiment of the present application;

[0036] Figure 2 is a composition diagram of a wind energy prediction model provided by an embodiment of the present application;

[0037] Figure 3 is a structural diagram of a wind energy prediction system based on multi-source data fusion provided by an embodiment of the present application;

[0038] Figure 4 is a hardware structure diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application and are not intended to limit the present application. When the following description refers to the accompanying drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with embodiments of the present application. They are only examples of systems and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0040] It can be understood that the terms "first", "second", and the like used in the present application can be used herein to describe various concepts, but unless specifically stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information. Depending on the context, the word "if" as used herein can be interpreted as "when" or "when" or "in response to determining".

[0041] The terms "at least one", "multiple", "each", "any" and the like used in the present application include one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0043] Accurate prediction of future wind energy data (such as wind power, wind speed, etc.) of a wind farm is crucial for stable and efficient operation of the power grid. For example, when the predicted future wind energy data is lower than the historical wind energy data, the power generation of a thermal power plant, etc. can be increased in advance, which is beneficial to ensure the overall stability of the grid power.

[0044] In the prior art, a machine learning model is usually used to analyze wind farm related historical data provided by a single data source to predict future wind energy data of the wind farm. However, this implementation has certain limitations. If the wind farm related historical data provided by a single data source is not accurate and comprehensive enough, the credibility of the future wind energy data of the wind farm predicted by the model is not high.

[0045] Therefore, the application provides a wind energy prediction method and system based on multi-source data fusion, an equipment and a medium.

[0046] The application provides a wind energy prediction method based on multi-source data fusion, relates to the field of wind power technology, can be applied to a terminal, can be applied to a server, and can be software running in the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, a vehicle-mounted terminal, and the like, but is not limited thereto. The server end can be configured as a stand-alone physical server, a server cluster composed of multiple physical servers, or a distributed system, and can also be configured as a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, big data, and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application for implementing the wind energy prediction method based on multi-source data fusion, but is not limited to the above forms.

[0047] The application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The application can also be practiced in a distributed computing environment in which tasks are performed by remote processing devices connected by a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0048] Figure 1 is an optional flowchart of the wind energy prediction method based on multi-source data fusion provided by the application, Figure 1 The method in the application can include, but is not limited to, steps S101 to S103:

[0049] Step S101, obtaining a plurality of meteorological data sets corresponding to a historical time period of a to-be-tested wind farm provided by a plurality of data sources;

[0050] Step S102, preprocessing the plurality of meteorological data sets to obtain a plurality of to-be-tested meteorological data sets corresponding thereto;

[0051] Step S103, inputting the plurality of to-be-tested meteorological data sets into a wind energy prediction model, the wind energy prediction model comprising an independent prediction module and a fusion prediction module, processing the plurality of to-be-tested meteorological data sets respectively by using the independent prediction module to obtain a plurality of initial prediction wind energy data sets corresponding to a future time period of the to-be-tested wind farm, and performing alignment splicing and analysis on the plurality of initial prediction wind energy data sets by using the fusion prediction module to obtain a final prediction wind energy data set of the to-be-tested wind farm in the future time period.

[0052] The steps S101 to S103 shown in the embodiments of the present application can realize accurate prediction of wind energy data of the to-be-tested wind farm in the future time period by using a specific wind energy prediction model to perform fusion analysis on the related meteorological data of the to-be-tested wind farm in the historical time period provided by the plurality of data sources.

[0053] In step S101 of some embodiments, the plurality of data sources can be understood as a plurality of different meteorological departments and research institutions, such as national meteorological service agencies, local meteorological stations, international meteorological organizations, and government-funded meteorological research centers, etc. In addition, the spatial coverage, spatial resolution, temporal resolution, and meteorological element types of the meteorological data sets provided by each data source can be different, and the meteorological element types include wind speed, wind direction, air temperature, air pressure, humidity, and turbulent kinetic energy, etc. For example, the meteorological data set provided by the first data source only contains two types of meteorological elements of air temperature and air pressure, and the meteorological data set provided by the second data source only contains three types of meteorological elements of air pressure, humidity, and turbulent kinetic energy.

[0054] In some embodiments, step S102 can include but is not limited to steps S201 to S202:

[0055] Step S201, performing normalization processing on the plurality of meteorological data sets to obtain a plurality of initial meteorological data sets corresponding thereto.

[0056] Specifically, when the plurality of meteorological data sets contain K types of meteorological elements, K is a positive integer greater than 1, for the i-th type of meteorological element, i = 1, 2,..., K, all values belonging to the i-th type of meteorological element are filtered out from the plurality of meteorological data sets, and then all filtered values are normalized by the following expression:

[0057]

[0058] wherein Z is a single value belonging to the i-th type of meteorological element, is a normalized result of the value Z, Z min is the minimum value among all values belonging to the i-th type of meteorological element, Z max is the maximum value among all values belonging to the i-th type of meteorological element;

[0059] For each type of meteorological element, all values belonging to the type of meteorological element contained in the plurality of meteorological data sets are normalized according to the above-mentioned embodiments, thereby obtaining a plurality of initial meteorological data sets after normalization.

[0060] Exemplarily, it is assumed that four types of meteorological elements, i.e., temperature, pressure, humidity and turbulent kinetic energy, are contained in the five meteorological data sets provided by the five data sources, it is assumed that the meteorological data set provided by the first data source is {A1, B1}, the meteorological data set provided by the second data source is {B2, C2, D2}, the meteorological data set provided by the third data source is {A3, B3, C3, D3}, the meteorological data set provided by the fourth data source is {A4, B4, C4}, and the meteorological data set provided by the fifth data source is {A5, B5, D5}, wherein A i represents temperature data, B i represents pressure data, C i represents humidity data, D i represents turbulent kinetic energy data, and each meteorological data contains corresponding meteorological values at different sampling time points within the historical time period; at this time, the data set {A1, A3, A4, A5} is normalized to obtain the data set {B1, B2, B3, B4, B5} is normalized to obtain the data set {C2, C3, C4} is normalized to obtain the data set {D2, D3, D5} is normalized to obtain and thereby the five initial meteorological data sets after normalization are respectively

[0061] Step S202, for each initial meteorological data set, the initial meteorological data set is divided according to a preset time interval, thereby obtaining a plurality of initial meteorological data subsets and forming a corresponding to-be-measured meteorological data set.

[0062] wherein the to-be-measured meteorological data set can be considered as a tensor with dimensions of (S k / L k , L k , W k , H k , D k ).k L is the length of the historical time period k S is the preset time interval, and the preset time interval is preferably set to be greater than or equal to the length of the future time period k L / S k W can be understood as the number of the plurality of initial meteorological data subsets k H k D is the width and height of the spatial grid formed by the plurality of initial meteorological data subsets k K is the number of categories of meteorological elements contained in the to-be-tested meteorological data set.

[0063] For example, assuming that the meteorological data set provided by the first data source is {A1, B1}, the meteorological data set provided by the second data source is {B2, C2, D2}, and the numerical collection intervals set by the first data source and the second data source are different, that is, the number of temperature values contained in the temperature data A1 and the number of pressure values contained in the pressure data B1 are both N1, the number of pressure values contained in the pressure data B2, the number of humidity values contained in the humidity data C2, and the number of turbulent kinetic energy values contained in the turbulent kinetic energy data D2 are all N2, but N1≠N2, N1 and N2 are both positive integers greater than 1, and each of the above meteorological values has a corresponding collection time in the historical time period; according to the preset time interval, the temperature data and the pressure data are synchronously divided along the time axis in the historical time period to obtain a plurality of divided temperature data and a plurality of divided pressure data Then, the single temperature data and the single pressure data of the same time period form an initial meteorological data subset corresponding to the first data source; similarly, according to the preset time interval, the pressure data humidity data and turbulent kinetic energy data are synchronously divided along the time axis in the historical time period to obtain a plurality of divided pressure data a plurality of divided humidity data and a plurality of divided turbulent kinetic energy data Then, the single pressure data single humidity data and single turbulent kinetic energy data of the same time period form an initial meteorological data subset corresponding to the second data source.

[0064] The steps S201 to S202 shown in the embodiments of the present application are beneficial to improving the calculation efficiency and prediction performance of the subsequent wind energy prediction model by normalizing and dividing the plurality of meteorological data sets.

[0065] In step S103 of some embodiments, referring to Figure 2 The wind energy prediction model includes an independent prediction module and a fusion prediction module, the independent prediction module includes a plurality of independent prediction sub-modules corresponding to a plurality of data sources, and the network structure of each independent prediction sub-module is the same. The internal implementation principle of a single independent prediction sub-module and the fusion prediction module is described below.

[0066] Specifically, for the independent prediction sub-module set for each data source, the independent prediction sub-module includes a spatial attention unit, a convolution unit and a long short-term memory network unit connected in sequence; when the independent prediction sub-module receives the to-be-tested meteorological data set associated with the data source, first, the spatial attention unit is used to perform preliminary feature extraction on the to-be-tested meteorological data set to obtain key feature data, second, the convolution unit is used to perform secondary feature extraction on the key feature data to obtain first deep feature data, and finally, the long short-term memory network unit is used to analyze the first deep feature data to obtain the initial prediction wind energy data set of the to-be-tested wind farm in the future time period. The initial prediction wind energy data set can be considered as a tensor with a dimension of (L k ,2), where 2 represents that the initial prediction wind energy data set includes initial prediction wind power data and initial prediction wind speed data with a time length of L k .

[0067] The spatial attention unit includes a first full connection layer and a spatial attention layer, and the to-be-tested meteorological data set is processed through the first full connection layer and the spatial attention layer in sequence to obtain the key feature data.

[0068] The spatial attention layer can use the following expression to implement the corresponding feature extraction operation:

[0069]

[0070] a 2 =g(a 1 )

[0071]

[0072] y=g(a 3 )

[0073] In the formula, y is the output data of the spatial attention layer, x is the input data of the spatial attention layer, * represents convolution operation, is the weight value of the convolution kernel with a size of 7x7, j=1,3, b is the offset value, f(·) is the activation function and is preferably the ReLU function, a 1 , a 2 and a3 are intermediate results of the spatial attention layer, g(·) represents a batch normalization operation, and σ(·) is a sigmoid function.

[0074] The convolution unit includes a first convolution layer, a first batch normalization layer, an average pooling layer, a maximum pooling layer, a first splicing layer, a second convolution layer, and a second batch normalization layer; the key feature data is sequentially processed by the first convolution layer and the first batch normalization layer to obtain first feature data; the first feature data is processed by the average pooling layer to obtain second feature data; the first feature data is processed by the maximum pooling layer to obtain third feature data; the second feature data and the third feature data are spliced by the first splicing layer to obtain fourth feature data; and the fourth feature data is sequentially processed by the second convolution layer and the second batch normalization layer to obtain the first deep feature data.

[0075] The LSTM (Long Short-Term Memory) network unit includes an LSTM (Long Short-Term Memory) layer, a Reshape layer, a second full connection layer, and a first ReLU (Rectified Linear Unit) layer; the first deep feature data is sequentially processed by the LSTM layer, the Reshape layer, the second full connection layer, and the first ReLU layer to obtain the initial predicted wind energy data set.

[0076] By using multiple independent prediction sub-modules to independently analyze and predict the heterogeneous and heterogeneous data collected by the corresponding multiple data sources, the spatial dimensions and data elements of the heterogeneous and heterogeneous data can be uniformly integrated, and the final output results of the multiple independent prediction sub-modules remain the same dimensions, which is conducive to the subsequent fusion prediction module for fast and effective data fusion analysis.

[0077] Specifically, the fusion prediction module comprises, in sequence, an alignment layer, a concatenation layer, a convolution layer, and a multi-layer perception; when the fusion prediction module receives the plurality of initial prediction wind energy data sets output by the independent prediction module, firstly, the plurality of initial prediction wind energy data sets are respectively time-interpolated according to the length of the future time period by the alignment layer to obtain a plurality of first prediction wind energy data sets corresponding thereto; secondly, the plurality of first prediction wind energy data sets are concatenated by the concatenation layer to obtain a second prediction wind energy data set after fusion; then, the second prediction wind energy data set is subjected to feature extraction by the convolution layer to obtain second deep feature data; finally, the second deep feature data is analyzed by the multi-layer perception to obtain a final prediction wind energy data set of the to-be-tested wind farm in the future time period, which can be considered as a tensor with a dimension of (L, 2), wherein 2 represents that the final prediction wind energy data set contains final prediction wind power data and final prediction wind speed data with a time length of L, and L is the length of the future time period.

[0078] Since each initial prediction wind energy data set output by the independent prediction module is a tensor with a dimension of (L, 2), by inputting each initial prediction wind energy data set into the alignment layer for processing, the corresponding first prediction wind energy data set becomes a tensor with a dimension of (L, 2), and the plurality of first prediction wind energy data sets contain the same number of wind energy values in the plurality of prediction wind energy data belonging to the same type. k ,2) of the independent prediction module, by inputting each initial prediction wind energy data set into the alignment layer for processing, the corresponding first prediction wind energy data set becomes a tensor with a dimension of (L, 2), and the plurality of first prediction wind energy data sets contain the same number of wind energy values in the plurality of prediction wind energy data belonging to the same type; then, by simultaneously processing the plurality of first prediction wind energy data sets through the concatenation layer, the second prediction wind energy data set becomes a tensor with a dimension of (L, 2K), K can be understood as the number of data sources; thereby, the time dimension of heterogeneous data can be uniformly integrated.

[0079] The multi-layer perception comprises a third fully connected layer, a fourth fully connected layer, a fifth fully connected layer, and a second ReLU layer, and the second deep feature data is processed in sequence through the third fully connected layer, the fourth fully connected layer, the fifth fully connected layer, and the second ReLU layer to obtain the final prediction wind energy data set.

[0080] By adopting the fusion prediction module, the plurality of initial prediction wind energy data sets provided by the independent prediction module can be effectively integrated, and the related wind energy data of the to-be-tested wind farm in the future time period can be more accurately predicted.

[0081] In step S103 of some embodiments, the wind energy prediction model is obtained by training an initial wind energy prediction model built in advance, and the corresponding training process can include, but is not limited to, steps S301 to S305:

[0082] Step S301, a plurality of wind energy data sets corresponding to a plurality of target wind farms in a first time period are acquired; wherein the length of the first time period is the same as the length of the future time period, but the first time period covers a past time range.

[0083] The wind energy data set of each target wind farm in the first time period includes wind power data and wind speed data of the target wind farm in the first time period, the wind power data includes wind power values of the target wind farm at different collection time points in the first time period, the wind power value refers to the power generation power of the target wind farm at a collection time point, and the wind speed data includes wind speed values of the target wind farm at different collection time points in the first time period; as an optional implementation, the wind energy data set of the target wind farm in the first time period is acquired by on-site monitoring, that is, wind speed sensors are installed at key positions such as meteorological towers or pole towers of the target wind farm to realize the collection of wind speed values at regular time points, and power meters are installed at the output ends of the wind power generation devices of the target wind farm to realize the collection of wind power values at regular time points, and the sampling time points of the wind speed sensors and the power meters are the same.

[0084] Step S302, the plurality of wind energy data sets are preprocessed to obtain a plurality of first wind energy data sets, and the corresponding preprocessing methods include wind energy data smoothing and wind power data normalization.

[0085] Step S303, a plurality of first meteorological data sets corresponding to a plurality of target wind farms in a second time period provided by a plurality of data sources are acquired; wherein the length of the second time period is the same as the length of the historical time period, the second time period is before the first time period, and the second time period is adjacent to the first time period.

[0086] Specifically, the number of the plurality of data sources is denoted as N, the number of the plurality of target wind farms is denoted as M, each data source can provide a first meteorological data set of each target wind farm in the second time period, that is, N data sources can provide N×M first meteorological data sets corresponding to M target wind farms in the second time period.

[0087] Step S304, a plurality of first meteorological data sets are preprocessed to obtain a plurality of second meteorological data sets, and the corresponding preprocessing methods include meteorological data normalization and meteorological data segmentation.

[0088] Step S305, the plurality of first wind energy data sets are used as training labels, and the initial wind energy prediction model is trained in combination with the plurality of second meteorological data sets to obtain the wind energy prediction model.

[0089] As an optional implementation, a mean square error function or a mean absolute error function or a Huber function is used to construct a loss function required for model training, and an Adam optimizer or an SGD optimizer is used to adjust the model parameters in the training process to minimize the loss function.

[0090] The steps S301 to S305 shown in the embodiments of the present application train the initially built wind energy prediction model by taking the plurality of wind energy data sets corresponding to the plurality of target wind farms in the first time period and the plurality of first meteorological data sets corresponding to the plurality of target wind farms in the second time period provided by the plurality of data sources as relevant training samples, so that the finally trained wind energy prediction model has better performance.

[0091] In some embodiments, step S302 can include but is not limited to steps S401 to S403:

[0092] Step S401, the plurality of wind energy data sets are stationary processed to obtain corresponding plurality of initial wind energy data sets.

[0093] As an optional implementation, an EMD (Empirical Mode Decomposition) algorithm is used to perform stationary processing on the wind power data and the wind speed data contained in each wind energy data set; taking the wind power data and the wind speed data contained in any one wind energy data set as an example, the wind power data is empirically mode decomposed to obtain a plurality of first sub-modes, and a given number of low-frequency first sub-modes are selected from the plurality of first sub-modes to form new wind power data, and the wind speed data is empirically mode decomposed to obtain a plurality of second sub-modes, and the given number of low-frequency second sub-modes are selected from the plurality of second sub-modes to form new wind speed data, and finally the new wind power data and the new wind speed data are combined to form the corresponding initial wind energy data set. The EMD algorithm can decompose any time series into a plurality of internal mode functions of different time scales and a long-term trend item, so that the new sequence formed after decomposition has stronger regularity compared with the original sequence.

[0094] Step S402, for each initial wind energy data set, the initial wind energy data set contains wind power data and wind speed data, according to the on-line capacity of the target wind farm associated with the initial wind energy data set, the wind power data is normalized by the following expression to obtain first wind power data:

[0095]

[0096] In the formula, cap is the on-line capacity of the target wind farm associated with the initial wind energy data set, P iFor the i th wind power value contained in the wind power data, it can be understood that it is obtained at the i th acquisition time in the first time period, For the normalized result of the wind power value P i .

[0097] Step S403, in combination with the wind speed data contained in the initial wind energy data set and the normalized first wind power data, a corresponding first wind energy data set is formed.

[0098] The steps S401 to S403 shown in the embodiments of the present application are beneficial to accelerating the convergence speed in the training process of the initial wind energy prediction model previously built by smoothing a plurality of wind energy data sets and normalizing the wind power data contained in each wind energy data set after smoothing.

[0099] In some embodiments, step S304 can include but is not limited to steps S501 to S502:

[0100] Step S501, normalizing a plurality of first meteorological data sets to obtain a plurality of corresponding first initial meteorological data sets;

[0101] Step S502, for each first initial meteorological data set, dividing the first initial meteorological data set according to the preset time interval to obtain a plurality of first initial meteorological data subsets to form a corresponding second meteorological data set.

[0102] It should be noted that the implementation of steps S501 to S502 is basically similar to the implementation of steps S201 to S202 described above, and will not be repeated here.

[0103] The steps S501 to S502 shown in the embodiments of the present application are beneficial to accelerating the convergence speed in the training process of the initial wind energy prediction model previously built by normalizing and dividing a plurality of first meteorological data sets.

[0104] Referring to Figure 3 , the embodiments of the present application also provide a wind energy prediction system based on multi-source data fusion, which can implement the wind energy prediction method based on multi-source data fusion described above, and the system comprises:

[0105] The acquisition module 601 is configured to acquire a plurality of meteorological data sets corresponding to the historical time period of the to-be-tested wind farm provided by a plurality of data sources;

[0106] The preprocessing module 602 is configured to preprocess the plurality of meteorological data sets to obtain a plurality of corresponding to-be-tested meteorological data sets;

[0107] The analysis module 603 is configured to input a plurality of to-be-tested meteorological data sets into a wind energy prediction model, the wind energy prediction model comprising an independent prediction module and a fusion prediction module, the plurality of to-be-tested meteorological data sets being processed by the independent prediction module to obtain a plurality of initial prediction wind energy data sets corresponding to the to-be-tested wind farm in a future time period, and the plurality of initial prediction wind energy data sets being aligned, spliced and analyzed by the fusion prediction module to obtain a final prediction wind energy data set of the to-be-tested wind farm in the future time period.

[0108] It can be understood that the contents in the above method embodiments are all applicable to the present system embodiments, the present system embodiments specifically implement the functions of the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0109] The present application also provides an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the wind energy prediction method based on multi-source data fusion described above when executing the computer program. The electronic device can include a tablet computer, a vehicle-mounted computer, or any other smart terminal.

[0110] It can be understood that the contents in the above method embodiments are all applicable to the present device embodiments, the present device embodiments specifically implement the functions of the above method embodiments, and achieve the same beneficial effects as the above method embodiments.

[0111] Please refer to Figure 4 , Figure 4 The electronic device of another embodiment is shown in the hardware structure, which comprises:

[0112] The processor 701 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits, and is configured to execute related programs to implement the technical solutions provided by the embodiments of the present application.

[0113] The memory 702 can be implemented by a ROM (Read-Only Memory), a static storage device, a dynamic storage device, or a RAM (Random Access Memory). The memory 702 can store an operating system and other application programs. When the technical solutions provided by the embodiments of the present application are implemented by software or firmware, the related program codes are stored in the memory 702 and called and executed by the processor 701 to implement the technical solutions provided by the embodiments of the present application.

[0114] The input / output interface 703 is configured to realize information input and output.

[0115] The communication interface 704 is configured to realize communication interaction between the device and other devices, and the communication can be realized through a wired manner (for example, a USB, a network cable, and the like) or a wireless manner (for example, a mobile network, WIFI, Bluetooth, and the like).

[0116] The bus 705 is configured to transmit information between various components (for example, the processor 701, the memory 702, the input / output interface 703, and the communication interface 704) of the device.

[0117] The processor 701, the memory 702, the input / output interface 703, and the communication interface 704 are connected to each other through the bus 705 to realize communication connection between the device.

[0118] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to realize the wind energy prediction method based on multi-source data fusion.

[0119] It can be understood that the contents in the above method embodiments are applicable to the storage medium embodiments, the storage medium embodiments specifically realize the functions of the above method embodiments, and the same beneficial effects as those of the above method embodiments are achieved.

[0120] The memory is a non-transitory computer readable storage medium, and can be used to store a non-transitory software program and a non-transitory computer executable program. In addition, the memory can include a high-speed random access memory, and can further include a non-transitory memory, for example, at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory can optionally include a memory remotely arranged relative to the processor, and the remote memory can be connected to the processor through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0121] The embodiments described in the embodiments of the present application are used to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. It can be understood by those skilled in the art that, with the evolution of technology and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0122] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and can include more or fewer steps than those shown in the figures, or combine certain steps, or different steps.

[0123] The system embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purposes of the embodiments.

[0124] Those skilled in the art can understand that all or some steps in the method disclosed above, and the functional modules / units in the system and the device can be implemented as software, firmware, hardware and appropriate combinations thereof.

[0125] The terms "first", "second", "third", "fourth" and the like in the description of the application and in the claims of the foregoing drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of the terms so construed can be interchanged, such that, for example, a description

[0126] It should be understood that in this application, "at least one" means one or more, and "multiple" means two or more. "And / or" is used to describe the relationship between the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean: only A, only B, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c, can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0127] In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other manners. For example, the system embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, and electrical, mechanical or other forms.

[0128] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0129] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0130] If the integrated unit is realized 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 solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program storage media.

[0131] The preferred embodiments of the present application are described above with reference to the accompanying drawings, but this does not limit the scope of the present application. Any modifications, equivalent replacements and improvements made by those skilled in the art without departing from the scope and spirit of the present application should be within the scope of the present application.

Claims

1. A wind energy prediction method based on multi-source data fusion, characterized in that, The method includes: Obtain multiple meteorological datasets corresponding to the wind farm under test within a historical time period from multiple data sources; The multiple meteorological datasets are preprocessed to obtain multiple corresponding meteorological datasets to be measured; The multiple meteorological datasets to be measured are input into the wind energy prediction model, which includes an independent forecast module and a fusion forecast module. The independent forecast module processes the multiple meteorological datasets to be measured separately to obtain multiple initial predicted wind energy datasets corresponding to the wind farm to be measured in the future time period. The fusion forecast module aligns, stitches and analyzes the multiple initial predicted wind energy datasets to obtain the final predicted wind energy dataset of the wind farm to be measured in the future time period. The independent forecasting module includes multiple independent forecasting sub-modules configured for the multiple data sources, and the network structure of each independent forecasting sub-module is the same. For each of the aforementioned data sources, the independent forecasting submodule includes a spatial attention unit, a convolutional unit, and a long short-term memory network unit. The spatial attention unit includes a first fully connected layer and a spatial attention layer. The convolutional unit includes a first convolutional layer, a first batch processing normalization layer, an average pooling layer, a max pooling layer, a first concatenation layer, a second convolutional layer, and a second batch processing normalization layer. The long short-term memory network unit includes an LSTM layer, a reshape layer, a second fully connected layer, and a first ReLU layer. When the independent forecast submodule receives the meteorological dataset to be measured associated with the data source, it processes the meteorological dataset to be measured sequentially through the first fully connected layer and the spatial attention layer to obtain key feature data. The key feature data is processed sequentially through the first convolutional layer and the first batch processing normalization layer to obtain first feature data; the first feature data is processed through the average pooling layer to obtain second feature data; the first feature data is processed through the max pooling layer to obtain third feature data; the second feature data and the third feature data are concatenated through the first concatenation layer to obtain fourth feature data; the fourth feature data is processed sequentially through the second convolutional layer and the second batch processing normalization layer to obtain first depth feature data. The first deep feature data is processed sequentially through the LSTM layer, the Reshape layer, the second fully connected layer, and the first ReLU layer to obtain the initial predicted wind energy dataset of the wind farm to be tested in the future time period.

2. The wind energy prediction method based on multi-source data fusion according to claim 1, characterized in that, The preprocessing of the multiple meteorological datasets to obtain the corresponding multiple meteorological datasets to be measured includes: The multiple meteorological datasets are normalized to obtain multiple corresponding initial meteorological datasets; For each initial meteorological dataset, the initial meteorological dataset is divided according to a preset time interval to obtain several initial meteorological data subsets to form a corresponding meteorological dataset to be measured; wherein, the preset time interval is greater than or equal to the length of the future time period.

3. The wind energy prediction method based on multi-source data fusion according to claim 1, characterized in that, The fusion prediction module includes an alignment layer, a stitching layer, a convolutional layer, and a multilayer perceptron; When the fusion forecast module receives the multiple initial predicted wind energy datasets, it uses the alignment layer to perform time interpolation on the multiple initial predicted wind energy datasets according to the length of the future time period to obtain multiple corresponding first predicted wind energy datasets; it uses the stitching layer to stitch the multiple first predicted wind energy datasets to obtain the fused second predicted wind energy dataset; and it uses the convolutional layer to extract features from the second predicted wind energy dataset to obtain second deep feature data. The second deep feature data is analyzed using the multilayer perceptron to obtain the final predicted wind energy dataset of the wind farm under test in the future time period.

4. The wind energy prediction method based on multi-source data fusion according to claim 2, characterized in that, The wind energy prediction model was trained in the following way: Obtain multiple wind energy datasets corresponding to multiple target wind farms within the first time period; The multiple wind energy datasets are preprocessed to obtain multiple corresponding first wind energy datasets; Obtain several first meteorological datasets corresponding to the multiple target wind farms provided by the multiple data sources within a second time period; wherein, the second time period is before the first time period, the length of the second time period is the same as the length of the historical time period, and the length of the first time period is the same as the length of the future time period; The aforementioned first meteorological datasets are preprocessed to obtain corresponding second meteorological datasets; The wind energy prediction model is obtained by using the multiple first wind energy datasets as training labels and combining them with the multiple second meteorological datasets to train the pre-built initial wind energy prediction model.

5. The wind energy prediction method based on multi-source data fusion according to claim 4, characterized in that, The preprocessing of the plurality of wind energy datasets to obtain the corresponding plurality of first wind energy datasets includes: The multiple wind energy datasets are stabilized to obtain multiple corresponding initial wind energy datasets; For each of the initial wind energy datasets, the initial wind energy datasets contain wind power data and wind speed data. Based on the operating capacity of the target wind farm associated with the initial wind energy datasets, the wind power data is normalized to obtain the first wind power data. By combining the wind speed data and the first wind power data, a corresponding first wind energy dataset is formed.

6. The wind energy prediction method based on multi-source data fusion according to claim 4, characterized in that, The preprocessing of the plurality of first meteorological datasets to obtain the corresponding plurality of second meteorological datasets includes: The aforementioned first meteorological datasets are normalized to obtain the corresponding first initial meteorological datasets; For each of the first initial meteorological datasets, the first initial meteorological datasets are divided according to the preset time interval to obtain several first initial meteorological data subsets to form the corresponding second meteorological datasets.

7. A wind energy prediction system based on multi-source data fusion, characterized in that, The system includes: The acquisition module is used to acquire multiple meteorological datasets corresponding to the wind farm under test within a historical time period from multiple data sources; The preprocessing module is used to preprocess the multiple meteorological datasets to obtain multiple corresponding meteorological datasets to be measured; An analysis module is used to input the multiple meteorological datasets to be measured into a wind energy prediction model. The wind energy prediction model includes an independent forecast module and a fusion forecast module. The independent forecast module processes the multiple meteorological datasets to be measured separately to obtain multiple initial predicted wind energy datasets corresponding to the wind farm to be measured in the future time period. The fusion forecast module aligns, stitches, and analyzes the multiple initial predicted wind energy datasets to obtain the final predicted wind energy dataset of the wind farm to be measured in the future time period. The independent forecasting module includes multiple independent forecasting sub-modules configured for the multiple data sources, and the network structure of each independent forecasting sub-module is the same. For each of the aforementioned data sources, the independent forecasting submodule includes a spatial attention unit, a convolutional unit, and a long short-term memory network unit. The spatial attention unit includes a first fully connected layer and a spatial attention layer. The convolutional unit includes a first convolutional layer, a first batch processing normalization layer, an average pooling layer, a max pooling layer, a first concatenation layer, a second convolutional layer, and a second batch processing normalization layer. The long short-term memory network unit includes an LSTM layer, a reshape layer, a second fully connected layer, and a first ReLU layer. When the independent forecast submodule receives the meteorological dataset to be measured associated with the data source, it processes the meteorological dataset to be measured sequentially through the first fully connected layer and the spatial attention layer to obtain key feature data. The key feature data is processed sequentially through the first convolutional layer and the first batch processing normalization layer to obtain first feature data; the first feature data is processed through the average pooling layer to obtain second feature data; the first feature data is processed through the max pooling layer to obtain third feature data; the second feature data and the third feature data are concatenated through the first concatenation layer to obtain fourth feature data; the fourth feature data is processed sequentially through the second convolutional layer and the second batch processing normalization layer to obtain first depth feature data. The first deep feature data is processed sequentially through the LSTM layer, the Reshape layer, the second fully connected layer, and the first ReLU layer to obtain the initial predicted wind energy dataset of the wind farm to be tested in the future time period.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images