High-precision wind-solar combined power prediction method

Abstract

The invention provides a high-precision wind-solar combined power prediction method. The method comprises the following steps: A, firstly, respectively collecting wind power generation data and photovoltaic power generation data; b, preprocessing the collected wind power generation data and photovoltaic power generation data; c, performing feature extraction on the preprocessed power generation data; d, introducing a time sequence attention mechanism according to the space-time correlation and the load curve of the wind-solar power generation power, and constructing a wind-solar power generation power joint prediction model based on an improved GRU-CNN algorithm by means of the time sequence feature mining capability of the GRU network and the space feature mining capability of the CNN network; e, carrying out the training, verification and testing of the model, and achieving the combined prediction of the wind-solar power generation power. And F, finally, on the basis of the high-precision meteorological data, the prediction accuracy performance of different methods is tested and evaluated, an optimal selection strategy is formed, and the optimal prediction method is obtained.The high-precision wind-solar combined power prediction method adopted by the invention can realize combined prediction of wind-solar power generation power, and the prediction precision is high.

Description

Technical Field

[0001] The present application relates to the technical field of wind-solar combined power prediction, and in particular to a high-precision wind-solar combined power prediction method. Background Art

[0003] The power of wind power and photovoltaic power is difficult to accurately predict due to its inherent uncertainty of random fluctuations. Random fluctuations in power may cause a serious imbalance between power generation and load, thereby causing large frequency deviations or voltage stability problems, and increasing system operation risks. In power systems with low inertia and limited frequency response capabilities, the uncertainty of wind power and photovoltaic power has a particularly significant negative impact on the system. When the power of grid-connected renewable energy fluctuates, the system needs to provide more flexible power sources to balance the fluctuations in wind power output, but the increase in the penetration rate of renewable energy will correspondingly reduce the proportion of flexible power sources, resulting in insufficient grid flexibility, which creates a contradiction between the output characteristics of renewable energy and the power system's requirements for power output. As the proportion of renewable energy installed capacity increases, this contradiction becomes increasingly severe and gradually becomes a bottleneck restricting the further development of the new energy industry.

[0004] The wind-solar complementary power generation system is one of the important implementation paths to deal with the volatility and intermittency of new energy. This project studies the complementary characteristics of wind and solar output, grasps the matching degree of wind and solar output in time and space, and establishes a wind-solar complementary integrated operation platform based on data-driven modeling technology and high-precision wind-solar power prediction technology to achieve economic and friendly dispatch of wind-solar complementary power generation systems, thereby smoothing out the fluctuations of new energy, improving power quality, and increasing green power consumption.

[0005] The wind-solar integrated station will become an important trend in the planning and construction of centralized renewable energy power stations in the future due to its advanced advantages in full utilization of land resources, complementary regulation of different power types, and compound utilization of transmission route equipment. In order to cope with this trend and corresponding needs, it is necessary to carry out research on high-precision prediction technology for the combined output curve of wind-solar stations in the ultra-short term, short term, and medium- to long-term.

[0006] The current forecasting technology can achieve the purpose of forecasting, but it cannot realize the joint forecasting of wind and solar power generation, and the forecasting accuracy is low. Summary of the invention

[0007] The present application provides a high-precision wind-solar combined power prediction method to solve the problem of low prediction accuracy in the prior art, and realizes the combined prediction of wind and solar power generation with high prediction accuracy.

[0008] The present application provides a high-precision wind-solar combined power prediction method, comprising the following steps:

[0009] A. First, collect wind power generation data and photovoltaic power generation data respectively;

[0010] B. Preprocess the collected wind power generation data and photovoltaic power generation data;

[0011] C. Extract features from the preprocessed power generation data;

[0012] D. Then, according to the spatiotemporal correlation of wind and solar power generation and the load curve, the temporal attention mechanism is introduced, and the joint prediction model of wind and solar power generation is constructed with the help of the temporal feature mining capability of the GRU network and the spatial feature mining capability of the CNN network;

[0013] E. Carry out model training, verification and testing to achieve joint prediction of wind and solar power generation;

[0014] F. Finally, based on high-precision meteorological data, the prediction accuracy performance of different methods is evaluated through testing, and a selection strategy is formed to obtain the best prediction method.

[0015] Preferably, the present application provides a high-precision wind-solar combined power prediction method, wherein the preprocessing method in step B is as follows:

[0016] a. Clean the collected power generation data;

[0017] b. Check the cleaned data to see if there are any missing data;

[0018] c. Correct the verified data;

[0019] d. Filtering the corrected data;

[0020] e. Update the filtered data.

[0021] Preferably, the present application provides a high-precision wind-solar combined power prediction method, wherein the data cleaning in step a includes time verification and sorting of the collected data, single-point threshold filtering of the data,

[0022] Preferably, the present application provides a high-precision wind-solar combined power prediction method, wherein the feature extraction method in step C is as follows:

[0023] a. Establish a data set, wherein the data set includes multiple sub-data sets to be extracted features;

[0024] b. Perform feature training on the data set to obtain a training model;

[0025] c. Extract the first keyword and the second keyword in the data set;

[0026] d. Circularly search each sub-data set in the data set, using the first keyword and the second keyword as initial conditions to search the sub-data set;

[0027] e. If the first keyword or the second keyword is matched in each sub-data set, the data is extracted.

[0028] Preferably, the present application provides a high-precision wind-solar combined power prediction method, wherein the joint prediction in step E adopts any one of the neural networks including BP neural network, LSTM neural network, bidirectional LSTM neural network, GRU neural network, bidirectional GRU neural network, CNN neural network, and extreme learning machine.

[0029] Preferably, the present application provides a high-precision wind-solar combined power prediction method, wherein in step D, an improved GRU-CNN algorithm is used to construct a joint prediction model.

[0030] Beneficial effects: The high-precision wind-solar joint power prediction method adopted by the present invention can realize the joint prediction of wind-solar power generation with high prediction accuracy; the present invention introduces a time-series attention mechanism based on the spatiotemporal correlation and load curve of wind-solar power generation, and uses the time-series feature mining capability of the GRU network and the spatial feature mining capability of the CNN network to construct a wind-solar power generation joint prediction model based on the improved GRU-CNN algorithm through the three core steps of heterogeneous information encoding, spatiotemporal feature fusion, and power prediction, and carries out model training, verification and testing to realize the joint prediction of wind-solar power generation. In addition, the feature extraction method adopted by the present invention can reduce the extraction difficulty, improve the feature extraction accuracy, and further improve the joint prediction accuracy by searching the first keyword and the second keyword.

[0031] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, a brief introduction will be given below to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 is a flow chart of the prediction method of the present invention;

[0034] Figure 2 This is a flow chart of the data preprocessing method of the present invention;

[0035] Figure 3This is a flow chart of the feature extraction method of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification, claims and drawings of this application are intended to cover non-exclusive inclusions.

[0038] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiments" in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0039] In addition, the terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order, and may explicitly or implicitly include one or more of the features.

[0040] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, the "connection" or "connection" of a mechanical structure may refer to a physical connection. For example, the physical connection may be a fixed connection, such as a fixed connection through a fixing member, such as a fixed connection through a screw, bolt or other fixing member; the physical connection may also be a detachable connection, such as a mutual snap-on or snap-fit ​​connection; the physical connection may also be an integral connection, such as a connection formed by welding, bonding or integral molding. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0041] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0042] Embodiment 1:

[0043] The present application provides a high-precision wind-solar combined power prediction method, comprising the following steps:

[0044] A. First, collect wind power generation data and photovoltaic power generation data respectively;

[0045] B. Preprocess the collected wind power generation data and photovoltaic power generation data;

[0046] C. Extract features from the preprocessed power generation data;

[0047] D. Then, according to the spatiotemporal correlation of wind and solar power generation and the load curve, the temporal attention mechanism is introduced, and the joint prediction model of wind and solar power generation is constructed with the help of the temporal feature mining capability of the GRU network and the spatial feature mining capability of the CNN network;

[0048] E. Carry out model training, verification and testing to achieve joint prediction of wind and solar power generation;

[0049] F. Finally, based on high-precision meteorological data, the prediction accuracy performance of different methods is evaluated through testing, and a selection strategy is formed to obtain the best prediction method.

[0050] In the present invention, the pretreatment method in step B is as follows:

[0051] a. Clean the collected power generation data;

[0052] b. Check the cleaned data to see if there are any missing data;

[0053] c. Correct the verified data;

[0054] d. Filtering the corrected data;

[0055] e. Update the filtered data.

[0056] The data cleaning in step a includes time verification and sorting of the collected data, and single-point threshold filtering of the data.

[0057] The data preprocessing method adopted by the present invention can achieve data sorting, noise reduction, and data filtering, thereby improving the efficiency of subsequent data processing.

[0058] In the present invention, the joint prediction in step E adopts any one of the neural networks including BP neural network, LSTM neural network, bidirectional LSTM neural network, GRU neural network, bidirectional GRU neural network, CNN neural network and extreme learning machine.

[0059] In the present invention, in step D, an improved GRU-CNN algorithm is used to construct a joint prediction model.

[0060] Embodiment 2:

[0061] A high-precision wind-solar combined power prediction method comprises the following steps:

[0062] A. First, collect wind power generation data and photovoltaic power generation data respectively;

[0063] B. Preprocess the collected wind power generation data and photovoltaic power generation data;

[0064] C. Extract features from the preprocessed power generation data;

[0065] D. Then, according to the spatiotemporal correlation of wind and solar power generation and the load curve, the temporal attention mechanism is introduced, and the joint prediction model of wind and solar power generation is constructed with the help of the temporal feature mining capability of the GRU network and the spatial feature mining capability of the CNN network;

[0066] E. Carry out model training, verification and testing to achieve joint prediction of wind and solar power generation;

[0067] F. Finally, based on high-precision meteorological data, the prediction accuracy performance of different methods is evaluated through testing, and a selection strategy is formed to obtain the best prediction method.

[0068] In the present invention, the pretreatment method in step B is as follows:

[0069] a. Clean the collected power generation data;

[0070] b. Check the cleaned data to see if there are any missing data;

[0071] c. Correct the verified data;

[0072] d. Filtering the corrected data;

[0073] e. Update the filtered data.

[0074] The data cleaning in step a includes time verification and sorting of the collected data, and single-point threshold filtering of the data.

[0075] The data preprocessing method adopted by the present invention can achieve data sorting, noise reduction, and data filtering, thereby improving the efficiency of subsequent data processing.

[0076] In the present invention, the joint prediction in step E adopts any one of the neural networks including BP neural network, LSTM neural network, bidirectional LSTM neural network, GRU neural network, bidirectional GRU neural network, CNN neural network and extreme learning machine.

[0077] In the present invention, in step D, an improved GRU-CNN algorithm is used to construct a joint prediction model.

[0078] This embodiment also discloses the feature extraction method in step C as follows:

[0079] a. Establish a data set, wherein the data set includes multiple sub-data sets to be extracted features;

[0080] b. Perform feature training on the data set to obtain a training model;

[0081] c. Extract the first keyword and the second keyword in the data set;

[0082] d. Circularly search each sub-data set in the data set, using the first keyword and the second keyword as initial conditions to search the sub-data set;

[0083] e. If the first keyword or the second keyword is matched in each sub-data set, the data is extracted.

[0084] In the feature extraction method adopted by the present invention, by searching the first keyword and the second keyword, the extraction difficulty can be reduced, the feature extraction accuracy is improved, and the joint prediction accuracy is further improved.

[0085] Project implementation plan: According to the project feasibility study report, each station currently contains a wind power prediction system, and a light power prediction system will be installed later. Considering the actual situation of the project and the subsequent wind-solar joint prediction on-site commissioning and optimization project, it is expected to adopt a prediction method that predicts the power of wind power and photovoltaic power separately, and then adds the prediction results. The specific prediction steps are as follows:

[0086] Data collection: First, it is necessary to collect historical data of wind and photovoltaic power generation systems, including meteorological data such as wind speed, wind direction, solar radiation, temperature, and corresponding power generation data.

[0087] Data cleaning and preprocessing: Clean and preprocess the collected data, including removing outliers, processing missing data, and processing time series.

[0088] Feature engineering: Extract relevant features based on domain knowledge and data characteristics, such as the changing trends and seasonality of wind speed and solar radiation.

[0089] Model selection: Choose a suitable model to build a power prediction model. You can consider using machine learning models such as support vector machine (SVM), random forest, or deep learning models such as recurrent neural network (RNN).

[0090] Model training: Use historical data to train the selected model, optimize model parameters, and verify the accuracy of the model.

[0091] Model evaluation: Use test data to evaluate the performance of the model and determine the predictive effect of the model.

[0092] Model deployment: Deploy the trained model to the actual system for future power prediction.

[0093] In summary, the high-precision wind-solar joint power prediction method adopted in the present invention can realize the joint prediction of wind and solar power generation with high prediction accuracy; according to the spatiotemporal correlation and load curve of wind and solar power generation, the present invention introduces the time series attention mechanism, and with the help of the time series feature mining ability of the GRU network and the spatial feature mining ability of the CNN network, through the three core steps of heterogeneous information encoding, spatiotemporal feature fusion, and power prediction, a wind and solar power generation joint prediction model based on the improved GRU-CNN algorithm is constructed, and the model is trained, verified and tested to realize the joint prediction of wind and solar power generation.

[0094] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A high-precision wind-solar combined power prediction method, characterized in that: The following steps are involved: A. First, collect wind power generation data and photovoltaic power generation data respectively; B. Preprocess the collected wind power generation data and photovoltaic power generation data; C. Extract features from the preprocessed power generation data; D. Then, according to the spatiotemporal correlation of wind and solar power generation and the load curve, the temporal attention mechanism is introduced, and the joint prediction model of wind and solar power generation is constructed with the help of the temporal feature mining capability of the GRU network and the spatial feature mining capability of the CNN network; E. Carry out model training, verification and testing to achieve joint prediction of wind and solar power generation; F. Finally, based on high-precision meteorological data, the prediction accuracy performance of different methods is evaluated through testing, and a selection strategy is formed to obtain the best prediction method.

2. A high-precision wind-solar combined power prediction method according to claim 1, characterized in that: The pretreatment method in step B is as follows: a. Clean the collected power generation data; b. Check the cleaned data to see if there are any missing data; c. Correct the verified data; d. Filtering the corrected data; e. Update the filtered data.

3. The high-precision wind-solar combined power prediction method according to claim 1 is characterized in that: The data cleaning in step a includes time verification and sorting of the collected data, and single-point threshold filtering of the data.

4. The high-precision wind-solar combined power prediction method according to claim 1 is characterized in that: The feature extraction method in step C is as follows: a. Establish a data set, wherein the data set includes multiple sub-data sets to be extracted features; b. Perform feature training on the data set to obtain a training model; c. Extract the first keyword and the second keyword in the data set; d. Circularly search each sub-data set in the data set, using the first keyword and the second keyword as initial conditions to search the sub-data set; e. If the first keyword or the second keyword is matched in each sub-data set, the data is extracted.

5. The high-precision wind-solar combined power prediction method according to claim 1 is characterized in that: The joint prediction in step E adopts any one of BP neural network, LSTM neural network, bidirectional LSTM neural network, GRU neural network, bidirectional GRU neural network, CNN neural network and extreme learning machine.

6. The high-precision wind-solar combined power prediction method according to claim 1 is characterized in that: In the step D, an improved GRU-CNN algorithm is used to construct a joint prediction model.

Images