Wind power plant group control system based on wind power prediction

Abstract

The invention discloses a wind power plant group control system based on wind power prediction, which relates to the technical field of wind power generation and comprises a group control region division module, a multi-source data acquisition module, a data analysis and processing module, a wind power prediction module and a wind power plant group control module. The group control area division module is used for dividing group control areas of the wind turbines according to the positions of the plurality of wind turbines; and the multi-source data acquisition module is electrically connected with the group control region division module. According to the wind power plant group control system provided by the invention, multiple data sources are integrated, and a prediction algorithm and a data processing technology are adopted, so that the accuracy and reliability of wind power prediction can be remarkably improved, and a wind power plant manager can know the future power generation potential of the wind power plant in advance, so that more reasonable scheduling and operation decisions are made; based on an accurate wind power prediction result, intelligent group control of each wind turbine generator in a wind power plant can be realized, and wind energy resources can be utilized to the maximum extent.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and specifically relates to a wind farm group control system based on wind power prediction. Background Technique

[0002] In the technical field of wind power generation, the operation efficiency and power generation capacity of a wind farm are directly related to the sustainable utilization of energy and economic benefits. However, traditional wind farm management methods often rely on manual experience and simple automation control, making it difficult to achieve precise control and optimal scheduling of each wind turbine in the wind farm. Especially when facing complex and variable wind resources and environmental conditions, traditional management methods often struggle to ensure the stable operation and efficient power generation of the wind farm. Currently, with the rapid development of intelligent technologies, the control of wind turbines has changed from single-unit control to group control. And current group control technologies in the industry all rely on the feedforward wind speed and wind direction data of lidar, and lidar will be installed at some positions in the wind farm. Since the wind speed monitored by lidar is at a distance of about 100 meters to 200 meters, which is much smaller than the distance between wind turbines, the control prediction parameters of group control are inaccurate. Lidar mainly uses lasers to monitor the wind speed and wind direction in the forward direction. In rainy and foggy days, the wind speed and wind direction monitored by lidar will deviate greatly from the actual situation, resulting in the inability to use the lidar data for group control. For this reason, we propose a wind farm group control system based on wind power prediction. Summary of the Invention To solve the above technical problems, a wind farm group control system based on wind power prediction is provided, and this technical solution solves the above problems.

[0003] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A wind farm group control system based on wind power prediction, comprising: a group control area division module, a multi-source data acquisition module, a data analysis and processing module, a wind power prediction module, and a wind farm group control module; The group control area division module is used to divide the group control area of wind turbines according to the positions of a number of wind turbines; The multi-source data acquisition module is electrically connected to the group control area division module, and the multi-source data acquisition module is used to acquire past and real-time weather data, past wind power and power generation data of the wind farm, and real-time sensing data; The data analysis and processing module is electrically connected to the multi-source data acquisition module, and the data analysis and processing module is used to perform preprocessing, feature extraction, and data division on the acquired data; The wind power prediction module is electrically connected to the data analysis and processing module, and the wind power prediction module is used to predict the wind speed and wind direction of different positions in the wind farm of the current group control area; The wind farm group control module is electrically connected to the wind power prediction module, and the wind farm group control module is used to perform group control on the motors in the wind farm according to the prediction results.

[0004] Preferably, the group control area division module includes: a geographic information analysis unit and a distance measurement unit; The geographic information analysis unit is used to obtain the geographic coordinates and altitude geographic information of the location where the wind turbine is located. By comprehensively analyzing the geographic information, it judges the relevance and distribution characteristics of different wind turbines at the geographic location level, and uses this as a basis to assist in dividing the group control area; The distance measurement unit is used to calculate the actual distance between each wind turbine.

[0005] Preferably, by comprehensively analyzing the geographic information, judging the relevance and distribution characteristics of different wind turbines at the geographic location level, and using this as a basis to assist in dividing the group control area specifically includes: Using the geographic information analysis unit to collect the geographic coordinates and altitude of the location where the wind turbine is located; Comprehensively analyze the geographic information and judge the relevance of different wind turbines at the geographic location level; Analyze the distribution characteristics of the wind turbines, and the distribution characteristics include: whether they are concentrated, dispersed, or distributed along specific geographic features; According to the relevance and distribution characteristics, divide the wind turbines with relevance and similar distribution characteristics into the same group control area.

[0006] Preferably, the formula for relevance is: In the formula, A represents wind turbine A, B represents wind turbine B, represents the coordinates of wind turbine A, represents the coordinates of wind turbine B, represents the Euclidean distance between wind turbine A and wind turbine B; Combined with the distance measurement results and geographic information analysis, set a distance threshold. When the distance between wind turbines is less than the threshold, it is considered that the two wind turbines are close in geographical location and are divided into the same group control area.

[0007] Preferably, the multi-source data acquisition module includes: a weather data acquisition unit, a wind power historical data collection unit, a power generation historical data collection unit, and a real-time sensing data acquisition unit; The weather data acquisition unit is used to obtain past and real-time weather data from the professional databases of meteorological departments, meteorological monitoring stations, and satellite meteorological data platforms; The wind power historical data collection unit is used to collect the past wind power data of the wind farm, and the wind power data includes: the average wind speed, maximum wind speed, and wind direction frequency distribution of each machine position in different time periods; The power generation historical data collection unit is used to collect the past power generation data of the wind farm. The power generation data includes: the power generation amount, power generation power and power generation efficiency of each wind turbine at different past time points; The real-time sensing data collection unit is used to obtain real-time data from various sensors installed in the wind farm. The sensors include: a wind speed sensor, a wind direction sensor, a temperature sensor and a vibration sensor.

[0008] Preferably, the data analysis and processing module includes: a data cleaning unit, a data standardization unit and a feature extraction unit; The data cleaning unit is used to remove noise, error values and duplicate data in the acquired data; The data standardization unit is used to unify data from different sources and different magnitudes to a standard scale; The feature extraction unit is used to extract key features related to wind power prediction and group control from the original data.

[0009] Preferably, the wind power prediction module includes: a prediction model construction unit, a model training unit and a prediction result output unit; The prediction model construction unit is used to establish a wind power prediction model; The model training unit is used to train the constructed prediction model, use the training set data to adjust the parameters of the model, so that the prediction result of the model is close to the actual situation. During the training process, some gradient descent methods are adopted to minimize the prediction error, and the feedback of the validation set is used to prevent the model from overfitting; The prediction result output unit is used to output the prediction results of wind speed and wind direction.

[0010] Preferably, the method steps for establishing a wind power prediction model are as follows: The data obtained from the multi-source data acquisition module is preprocessed in the data analysis and processing module; Remove noise, error values and duplicate data in the data, and perform standardization processing on the data; Extract key features related to wind power prediction. The features include: the mean value, variance, wind direction change rate, temperature and air pressure of wind speed in different time periods; Divide the processed data into a training set and a test set; Based on the extracted feature data, construct a wind power prediction model, input the training set data into the wind power prediction model, evaluate the performance of the model through the validation set, and monitor the value of the loss function; Apply the trained wind power prediction model to the wind speed and wind direction prediction of the actual wind farm.

[0011] Preferably, the expression of the wind power prediction model is: The expression for updating the hidden layer state is: For each time step t, the hidden layer state is updated based on the input at the current time step and the hidden layer state at the previous time step , assuming the number of hidden layer nodes is , and the number of input features is , then: In the formula, represents the state of the j-th node in the hidden layer at time t, represents the feature of the i-th node in the hidden layer at time t, represents the element in the j-th row and i-th column of the weight matrix input to the hidden layer, represents the element in the j-th row and k-th column of the weight matrix from the hidden layer to the hidden layer, represents the j-th element of the bias vector of the hidden layer, represents the activation function; The calculation expression of the output layer is: Assuming the number of output layer nodes is , then the output is calculated as: In the formula, represents the l-th element of the output vector at time t, represents the element in the l-th row and j-th column of the weight matrix from the hidden layer to the output layer, represents the l-th element of the bias vector of the hidden layer.

[0012] Preferably, the wind farm group control module includes: an instruction generation unit and a status monitoring unit; The instruction generation unit is used to generate control instructions according to the results of the wind power prediction module; The status monitoring unit is used to monitor the operating status of each wind turbine in the wind farm in real time.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The wind farm group control system proposed by the present invention can significantly improve the accuracy and reliability of wind power prediction by integrating various data sources and adopting prediction algorithms and data processing technologies. It helps wind farm managers understand the future power generation potential of the wind farm in advance, so as to make more reasonable scheduling and operation decisions. Based on the accurate wind power prediction results, it can realize the intelligent group control of each wind turbine in the wind farm. By dynamically adjusting the operating state and output power of the wind turbines, it can maximize the utilization of wind energy resources, improve the overall power generation efficiency and stability of the wind farm. At the same time, the system can also flexibly adjust the group control strategy according to real-time data to meet the requirements under different environments and working conditions. Brief Description of the Drawings

[0014] Figure 1 It is a module connection diagram of a wind farm group control system based on wind power prediction. Detailed Embodiments

[0015] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0016] Referring to Figure 1 As shown, a wind farm group control system based on wind power prediction includes: a group control area division module, a multi-source data acquisition module, a data analysis and processing module, a wind power prediction module, and a wind farm group control module; The group control area division module plays a crucial role in the entire system. It divides the group control area of wind turbines according to the positions of several wind turbines. Among them, the geographical information analysis unit will obtain the geographical coordinates and altitude of the positions where the wind turbines are located. By comprehensively analyzing these geographical information, it can judge the relevance and distribution characteristics of different wind turbines at the geographical location level. This analysis result will be used as an important basis to assist in the division of the group control area. The distance measurement unit is responsible for calculating the actual distances between each wind turbine. By accurately measuring the distances, it can better understand the relative position relationship between the wind turbines and provide more accurate data support for subsequent group control decisions.

[0017] The multi-source data acquisition module is electrically connected to the group control area division module. Its function is to acquire past and real-time weather data, past wind power and power generation data of the wind farm, and real-time sensing data. This module contains multiple units. The weather data acquisition unit acquires past and real-time weather data from the professional databases of meteorological departments, meteorological monitoring stations, and satellite meteorological data platforms. These weather data are crucial for wind power prediction because weather conditions directly affect the power generation capacity of the wind farm. The wind power historical data collection unit is responsible for collecting past wind power data of the wind farm, including the average wind speed, maximum wind speed, and wind direction frequency distribution of each wind turbine position at different time periods. These data can help analyze the wind power characteristics of the wind farm and provide historical data support for the wind power prediction model. The power generation historical data collection unit collects past power generation data of the wind farm, such as the power generation amount, power generation power, and power generation efficiency of each wind turbine at different past time points. These data reflect the power generation performance of the wind farm and have important reference value for optimizing the group control strategy. The real-time sensing data acquisition unit acquires real-time data from various sensors installed in the wind farm. These sensors include wind speed sensors, wind direction sensors, temperature sensors, and vibration sensors. Real-time sensing data can timely reflect the current operating state of the wind farm and provide a basis for the real-time control of the system.

[0018] The data analysis and processing module is electrically connected to the multi-source data acquisition module. It is responsible for preprocessing, feature extraction, and data division of the acquired data. The data cleaning unit removes noise, error values, and duplicate data from the acquired data to ensure the accuracy and reliability of the data. The data standardization unit unifies data from different sources and different magnitudes to a standard scale for subsequent analysis and processing. The feature extraction unit extracts key features related to wind power prediction and group control from the original data. By extracting key features, the dimension of the data can be reduced, the analysis efficiency can be improved, and at the same time, it is helpful to improve the accuracy of wind power prediction.

[0019] The wind power prediction module is electrically connected to the data analysis and processing module. Its main function is to predict the wind speed and wind direction of different wind turbine positions in the current group control area of the wind farm. The prediction model construction unit establishes a wind power prediction model, which can predict future wind speed and wind direction based on historical data and real-time data. The model training unit trains the constructed prediction model, uses the training set data to adjust the parameters of the model, and makes the prediction results of the model close to the actual situation. During the training process, some gradient descent methods are used to minimize the prediction error, and the feedback of the validation set is used to prevent the model from overfitting. The prediction result output unit outputs the prediction results of wind speed and wind direction, providing a decision-making basis for the wind farm group control module.

[0020] The wind farm group control module is electrically connected to the wind power prediction module. This module is used to perform group control on the motors in the wind farm according to the prediction results. The instruction generation unit generates control instructions based on the results of the wind power prediction module. These instructions can adjust the operating state and output power of the wind turbines to achieve the goal of maximizing the utilization of wind energy resources. The status monitoring unit monitors the operating state of each wind turbine in the wind farm in real time to ensure that the wind turbines operate in a safe and stable state. At the same time, the status monitoring unit can also feedback the operating state of the wind turbines to other modules to adjust the group control strategy in a timely manner.

[0021] By comprehensively analyzing geographical information, judge the relevance and distribution characteristics of different wind turbines at the geographical location level, and use this as a basis to assist in dividing the group control area, specifically including: Use the geographical information analysis unit to comprehensively collect the geographical coordinates and altitude of the locations where the wind turbines are located. These geographical coordinates can accurately determine the specific location of each wind turbine on the earth, while the altitude reflects the terrain height situation where the wind turbine is located. By collecting this information, it provides basic data for subsequent comprehensive analysis.

[0022] Carry out the work of comprehensively analyzing geographical information. In this process, carefully judge the relevance of different wind turbines at the geographical location level. This relevance is reflected in many aspects. For example, wind turbines located on the same mountain range trend have a certain relevance in terms of wind direction and wind speed; or wind turbines located near the same river basin are affected by similar climates and terrains, and thus have relevance in terms of geographical location. By deeply analyzing these relevances, it is possible to better understand the mutual relationship between wind turbines and provide important clues for the division of the group control area.

[0023] Analyze the distribution characteristics of the wind turbines. These distribution characteristics include but are not limited to whether they are concentrated, dispersed, or distributed along specific geographical features. If the wind turbines are concentrated, it means that the areas where these wind turbines are located have similar wind resource conditions and topographies, and can be divided into a group control area for unified management and control. If they are dispersed, it is necessary to consider grouping together wind turbines that are relatively close or have a certain relevance according to specific circumstances for group control. When the wind turbines are distributed along specific geographical features, such as the coastline or the edge of the mountain range, these geographical features will have specific impacts on the operation of the wind turbines, so these factors need to be fully considered when dividing the group control area; According to the relevance and distribution characteristics, divide the wind turbines with relevance and similar distribution characteristics into the same group control area, which can ensure that the wind turbines in the same group control area have a certain similarity in geographical location, thus facilitating the adoption of unified control strategies and management methods.

[0024] The calculation formula for relevance is: In the formula, A represents wind turbine A, and B represents wind turbine B. represents the coordinates of wind turbine A. represents the coordinates of wind turbine B. represents the Euclidean distance between wind turbine A and wind turbine B. Meanwhile, combining the distance measurement results and geographical information analysis, a reasonable distance threshold is set. When the distance between wind turbines is less than this threshold, it is considered that the two wind turbines are geographically close and can be divided into the same group control area. In this way, the division of the group control area can be further optimized, and the management efficiency and operation stability of the wind farm can be improved.

[0025] The method steps for establishing a wind power prediction model are as follows: The data obtained from the multi-source data acquisition module will be sent to the data analysis and processing module for a series of preprocessing operations. This process is crucial because the original data has various problems that will affect the accuracy and reliability of the subsequent model.

[0026] In the preprocessing stage, several important tasks will be carried out. First, the noise in the data is removed. Noise is generated due to sensor errors and data transmission interference, which will make the data inaccurate and affect the prediction effect of the model. By using specific filtering algorithms or data smoothing techniques, the noise can be effectively removed. Second, the error values are removed. Error values are generated due to errors in the data acquisition process and equipment failures. These error values will have a serious negative impact on the training and prediction of the model, so they must be identified and removed. The error values can be removed by setting reasonable data ranges and performing data verification methods. Third, the duplicate data is removed. Duplicate data not only wastes computing resources but also causes overfitting of the model. By de-duplicating the data, the data quality and the performance of the model can be improved. At the same time, the data will also be standardized. Since the data from different sources have different magnitudes and units, the standardized processing can unify these data to a standard scale, facilitating subsequent analysis and processing.

[0027] Extract key features related to wind power prediction from the preprocessed data. These features are crucial for accurate wind power prediction. They include the mean wind speed over different time periods, which reflects the average speed of the wind over a period of time and is of great significance for wind power prediction. The wind speed variance can reflect the fluctuation of the wind speed. A larger variance means the instability of the wind, which will affect the operation and power output of the wind farm. The wind direction change rate is also an important feature. The change in wind direction will directly affect the power generation efficiency of wind turbines. In addition, temperature and air pressure also affect wind power. The change in temperature affects the air density, thereby affecting the energy of the wind. The change in air pressure affects the flow direction and speed of the wind. By extracting these key features, the dimensionality of the data can be reduced, and the training efficiency and prediction accuracy of the model can be improved.

[0028] Divide the processed data into a training set and a test set. The training set is used to train the wind power prediction model, and the test set is used to evaluate the performance of the model. The ratio of dividing the training set and the test set usually depends on the specific situation. Generally speaking, the proportion of the training set will be larger so that the model can fully learn the characteristics and laws of the data. When dividing the data, it is necessary to ensure the randomness and representativeness of the data to avoid data deviation.

[0029] Build a wind power prediction model based on the extracted feature data. Input the training set data into the wind power prediction model, and by continuously adjusting the parameters of the model, the model can better fit the training data. During the training process, it is necessary to evaluate the performance of the model through the validation set. The validation set is a part of the data divided from the training set, which is used to monitor the training process of the model to prevent overfitting of the model. Evaluate the performance of the model by calculating the loss function value and accuracy index of the model on the validation set. If the performance of the model on the validation set is not good, it is necessary to adjust the structure, parameters of the model or increase the amount of data to improve the model.

[0030] Apply the trained wind power prediction model to the wind speed and wind direction prediction of an actual wind farm. In actual applications, it is necessary to continuously collect new data and update and optimize the model to improve the prediction accuracy and adaptability of the model. At the same time, it is also necessary to analyze and evaluate the prediction results of the model in order to detect problems in time and take corresponding measures.

[0031] The expression of the wind power prediction model is: Among them, the expression for updating the hidden layer state is: For each time step t, the hidden layer state is updated based on the input at the current time step and the hidden layer state at the previous time step. Let the number of hidden layer nodes be and the number of input features be , then: In the formula, represents the state of the j-th node in the hidden layer at time t, represents the feature of the i-th node in the hidden layer at time t, represents the weight matrix input to the hidden layer the element in the i-th column of the j-th row in represents the weight matrix from the hidden layer to the hidden layer the element in the k-th column of the j-th row in represents the bias vector of the hidden layer the j-th element of represents the activation function; Among them, the calculation expression of the output layer is: Suppose the number of output layer nodes is , then the output The calculation formula of is: In the formula, represents the l-th element of the output vector at time t, represents the weight matrix from the hidden layer to the output layer the element in the j-th column of the l-th row in represents the bias vector of the hidden layer the l-th element of

[0032] The usage process of the present invention is as follows: The system will comprehensively collect historical and real-time data of the wind farm and its surrounding environment, including weather conditions, wind power historical data, and real-time sensing data; based on the collected geographical information, the system will automatically divide the wind farm into regions for more refined management of the wind turbines in each region; preprocess the collected data to remove noise and error values, and extract key features related to wind power prediction; using the constructed wind power prediction model, the system will accurately predict the wind speed and wind direction in the current and future periods; according to the prediction results, the system will automatically generate control instructions for each unit in the wind farm and monitor the execution effect in real time to ensure the efficient and stable operation of the wind farm.

[0033] In summary, the advantages of the present invention are as follows: The wind farm group control system of the present invention integrates multiple data sources, uses prediction algorithms and data processing technologies to improve the accuracy and reliability of wind power prediction, and helps managers understand the power generation potential in advance to make reasonable decisions. Based on accurate prediction, intelligent group control of wind turbines can be realized, the operating state and output power can be dynamically adjusted, the power generation efficiency and stability can be improved, and the group control strategy can be flexibly adjusted according to real-time data to meet different needs.

[0034] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, various changes and improvements will occur to the present invention, and these changes and improvements fall within the scope of the present invention claimed. The scope of protection required by the present invention is defined by the appended claims and their equivalents.

Claims

1. A wind farm group control system based on wind power prediction, characterized in that: include: Group control area division module, multi-source data acquisition module, data analysis and processing module, wind power prediction module, wind farm group control module; The group control area division module is used to divide the group control area of ​​the wind turbine according to the positions of several wind turbines; A multi-source data acquisition module is electrically connected to the group control area division module, and the multi-source data acquisition module is used to acquire past and real-time weather data, past wind power and power generation data of the wind farm, and real-time sensor data; The data analysis and processing module is electrically connected to the multi-source data acquisition module, and the data analysis and processing module is used to perform preprocessing, feature extraction and data division on the acquired data; The wind power prediction module is electrically connected to the data analysis and processing module, and the wind power prediction module is used to predict the wind speed and wind direction of different positions of the wind farm in the current group control area; The wind farm group control module is electrically connected to the wind power prediction module, and the wind farm group control module is used to perform group control on the motors in the wind farm according to the prediction results.

2. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The group control area division module includes: geographic information analysis unit and distance measurement unit; The geographic information analysis unit is used to obtain the geographic coordinates and altitude information of the wind turbine location. Through comprehensive analysis of the geographic information, the correlation and distribution characteristics of different wind turbines at the geographic location level are determined, and based on this, the group control area is divided. The distance measurement unit is used to calculate the actual distance between each wind turbine.

3. A wind farm group control system based on wind power prediction according to claim 2, characterized in that: Through comprehensive analysis of geographic information, the correlation and distribution characteristics of different wind turbines at the geographical location level are judged, and based on this, the group control areas are divided into: Using a geographic information analysis unit to collect the geographic coordinates and altitude of the wind turbine location; Comprehensively analyze geographic information to determine the correlation between different wind turbines at the geographical location level; Analyze the distribution characteristics of wind turbines, including whether they are concentrated, dispersed, or distributed along specific geographical features; According to the correlation and distribution characteristics, wind turbines with correlation and similar distribution characteristics are divided into the same group control area.

4. A wind farm group control system based on wind power prediction according to claim 3, characterized in that: The calculation formula of correlation is: In the formula, A represents wind turbine A, B represents wind turbine B, represents the coordinates of wind turbine A, represents the coordinates of wind turbine B, Represents the Euclidean distance between wind turbine A and wind turbine B; Combining the distance measurement results and geographic information analysis, a distance threshold is set. When the distance between wind turbines is less than the threshold, the two wind turbines are considered to be geographically close and are divided into the same group control area.

5. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The multi-source data acquisition module includes: a weather data acquisition unit, a wind power history data collection unit, a power generation history data collection unit and a real-time sensor data collection unit; The weather data collection unit is used to obtain past and real-time weather data from the meteorological department's professional database, meteorological monitoring stations, and satellite meteorological data platforms; The wind power historical data collection unit is used to collect the past wind power data of the wind farm, and the wind power data includes: the average wind speed, maximum wind speed and wind direction frequency distribution of each position in different time periods; The power generation history data collection unit is used to collect the past power generation data of the wind farm, and the power generation data includes: the power generation, power generation and power generation efficiency of each wind turbine at different time points in the past; The real-time sensing data acquisition unit is used to obtain real-time data from various sensors installed in the wind farm, and the sensors include: wind speed sensor, wind direction sensor, temperature sensor and vibration sensor.

6. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The data analysis and processing module includes: data cleaning unit, data standardization unit and feature extraction unit; The data cleaning unit is used to remove noise, error values ​​and duplicate data from the acquired data; Data standardization units are used to unify data from different sources and of different magnitudes into a standard scale; The feature extraction unit is used to extract key features related to wind power prediction and group control from the original data.

7. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The wind power prediction module includes: a prediction model building unit, a model training unit and a prediction result output unit; The prediction model building unit is used to build a wind power prediction model; The model training unit is used to train the constructed prediction model. The training set data is used to adjust the model parameters so that the model's prediction results are close to the actual situation. During the training process, some gradient descent methods are used to minimize the prediction error, and the feedback from the validation set is used to prevent the model from overfitting. The prediction result output unit is used to output the prediction results of wind speed and wind direction.

8. A wind farm group control system based on wind power prediction according to claim 7, characterized in that: The method steps for establishing a wind power prediction model are as follows: The data acquired from the multi-source data acquisition module is pre-processed in the data analysis and processing module; Remove noise, error values ​​and duplicate data from the data and standardize the data; Extracting key features related to wind power prediction, including: wind speed mean, variance, wind direction change rate, temperature and air pressure in different time periods; Divide the processed data into training set and test set; Build a wind power prediction model based on the extracted feature data, input the training set data into the wind power prediction model, evaluate the performance of the model on the validation set, and monitor the value of the loss function; The trained wind power prediction model is applied to the wind speed and direction prediction of the actual wind farm.

9. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The expression of wind power prediction model is: The hidden layer state update expression is: For each time step t, the hidden layer state The update is based on the input of the current time step and the hidden layer state at the previous time step , let the number of hidden layer nodes be , the number of input features is ,but: In the formula, represents the state of the jth node in the hidden layer at time t, represents the feature of the i-th node in the hidden layer at time t, Represents the input to the hidden layer weight matrix The element in the jth row and ith column of Represents the hidden layer to hidden layer weight matrix The element in row j and column k of Represents the hidden layer bias vector The jth element of represents the activation function; The output layer calculation expression is: Assume the number of output layer nodes is , then the output The calculation formula is: In the formula, represents the lth element of the output vector at time t, Represents the weight matrix from the hidden layer to the output layer The element in row l and column j in Represents the hidden layer bias vector The lth element of .

10. A wind farm group control system based on wind power prediction according to claim 1, characterized in that: The wind farm group control module includes: a command generation unit and a status monitoring unit; The instruction generation unit is used to generate control instructions according to the results of the wind power prediction module; The status monitoring unit is used to monitor the operating status of each wind turbine in the wind farm in real time.

Images