A wind power prediction system and method based on solar terms

Abstract

The present invention discloses a wind power prediction system and method based on solar terms, which relates to the field of wind power technology and includes: a data acquisition module, the data acquisition module is used to collect historical data and environmental data of the current environment; a data calibration module; a real-time prediction module, the real-time prediction module is used to predict the power of a wind farm in a combined manner according to the data calibrated by the data calibration module; a cluster prediction module, the cluster prediction module is used to predict the power obtained by the real-time prediction module; by providing the data calibration module and the real-time prediction module, the data obtained by the sensor can be calibrated according to the historical data to reduce the error caused by the measurement, and the predicted power obtained by different methods is combined and calculated to obtain the final predicted power value, so as to avoid the situation that some prediction models may perform well under specific conditions but have large actual prediction errors.

Description

Technical Field

[0001] The present invention relates to the field of wind power technology, and more specifically, to a solar term-based wind power prediction system and method. Background Art

[0002] The wind power prediction system is an innovative method that uses the 24 solar terms in traditional Chinese culture to improve the accuracy of wind power prediction. This method takes into account the impact of seasonal and climate change on wind speed, wind direction and other wind farm operating conditions. By dividing meteorological data samples according to solar terms and establishing a prediction model corresponding to each solar term, the impact of seasonal and climate change is reduced to a certain extent, and the prediction accuracy is improved.

[0003] However, existing prediction systems usually rely on historical data and numerical weather forecasts, which may be missing or inaccurate, affecting prediction accuracy; some prediction models may perform well under specific conditions, but lack generalization capabilities under other conditions, resulting in large prediction errors; the system may fail to fully utilize real-time data for corrections and updates, resulting in inaccurate prediction results; reliance on a single prediction model may make the prediction results sensitive to specific errors or abnormal conditions and lack flexibility; for power prediction of wind power clusters, it is necessary to consider the interaction of multiple wind farms and overall meteorological conditions, which makes modeling difficult. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a wind power prediction system and method based on solar terms.

[0005] In order to achieve the above object, on the one hand, the present invention adopts the following technical solutions:

[0006] A wind power prediction system based on solar terms, comprising:

[0007] A data collection module, wherein the data collection module is used to collect historical data and environmental data of the current environment;

[0008] A data calibration module, which is used to calibrate the information acquired by the data acquisition module to reduce data errors;

[0009] A real-time prediction module is used to perform power prediction on a wind farm according to the data calibrated by the data calibration module in a combined manner, specifically:

[0010] Obtain the physical predicted power value of the wind farm and mark it as Q;

[0011] Obtain the wind farm's time series predicted power value and mark it as W;

[0012] Obtain the power value predicted by the artificial intelligence method of the wind farm and mark it as E;

[0013] According to the formula The final predicted power value R is calculated and obtained, where , and is the preset proportional coefficient;

[0014] A threshold value of the predicted power value is set in advance and marked as T. The predicted power value R is compared with the threshold value T. If the predicted power value R is less than the threshold value T, it is judged that the power generation power of the wind farm is insufficient. If not, it is judged that the wind farm is operating normally.

[0015] The cluster prediction module is used to further predict the predicted power of the entire wind farm cluster by integrating the interaction of multiple wind farms and the overall meteorological conditions.

[0016] Preferably, the physical method predicted power value of the wind farm is obtained as follows:

[0017] The efficiency value of the wind turbine generator set in the wind farm is obtained and marked as

[0018] According to the formula , calculating and obtaining the physical method predicted power value of the wind farm;

[0019] in is the air density value, d is the radius of the wind wheel in the wind power equipment, and V is the wind speed.

[0020] Preferably, the wind farm power value predicted by the time series method is obtained in the following manner, specifically:

[0021] Get the predicted power value one year before the current time point and mark it as

[0022] According to the formula , calculate and obtain the time series method predicted power value of the wind farm ;

[0023] Among them, b is the preset smoothing parameter.

[0024] Preferably, the method for obtaining the power value predicted by the artificial intelligence method of the wind farm is as follows, specifically:

[0025] According to the formula , calculate and obtain the artificial intelligence predicted power value E of the wind farm;

[0026] in is the activation function, is the input feature vector, which is any one of wind speed, air density, air pressure, temperature and humidity. is the weight corresponding to each feature vector, n is the number of input features, which is 5, and b is the bias term.

[0027] Preferably, the specific working mode of the data calibration module is as follows, specifically:

[0028] Get the real-time environmental information collected by the data acquisition module and mark it as U;

[0029] The environmental information is any one of wind speed, air density, air pressure, temperature and humidity;

[0030] According to the formula , calculate and obtain the calibrated environmental information Y, where The default value of the environment information, is the preset correction factor.

[0031] Preferably, the preset value of the environmental information The method of obtaining is as follows:

[0032] Get the node dates of the solar terms in the five years before the current time period, divide the environmental information obtained according to the time nodes of the 24 solar terms, establish 24 groups of data sample sets corresponding to the solar terms, get the average of the environmental information data recorded in the data sample set, and mark it as the preset value .

[0033] Preferably, the preset correction coefficient The way to obtain is as follows:

[0034] Get each detected environment information and preset value in the year before the current time period Deviation;

[0035] Get the average of all deviations and mark it as P;

[0036] The standard deviation of all deviations is calculated and marked as

[0037] According to the formula , calculate and obtain the preset correction coefficient .

[0038] Preferably, the specific working mode of the cluster prediction module is as follows, specifically:

[0039] Taking the wind farm as the center and the preset radius as the center, a circle is drawn to obtain the cluster area;

[0040] Obtain the difference between the predicted power value R and the threshold value T of all wind farms in the cluster area;

[0041] A difference threshold is set in advance, all differences are compared with the threshold, all wind farms with differences less than the threshold are marked as benchmark wind farms, and the number of benchmark wind farms is marked as K.

[0042] Preferably, the total installed capacity of the wind farms in the cluster area is obtained and marked as H;

[0043] According to the formula , calculate and obtain the total predicted power L of the cluster area,

[0044] in is the power forecast value of the i-th benchmark wind farm, is the installed capacity of the ith benchmark wind farm.

[0045] On the other hand, the present invention also proposes a wind power prediction method based on solar terms, comprising the following steps:

[0046] Step 1: Collect historical data and numerical weather forecasts, and calibrate the data to reduce data errors;

[0047] Step 2: Use the calibrated data to predict the power of the wind farm and determine whether the power of the wind farm meets expectations;

[0048] Step 3: Based on the obtained predicted power, the predicted power of the entire wind farm cluster is further predicted by combining the interaction of multiple wind farms and the overall meteorological conditions.

[0049] Beneficial effects: By setting up a data calibration module and a real-time prediction module, the data obtained by the sensor can be calibrated according to historical data to reduce the error caused by the measurement. The predicted power obtained by different methods is combined and calculated to obtain the final predicted power value, avoiding the situation where some prediction models may perform well under specific conditions, but have insufficient generalization ability under other conditions, resulting in large prediction errors;

[0050] By providing a cluster prediction module, the interaction between multiple wind farms can be considered to obtain the total predicted power of the entire wind farm cluster, thereby avoiding large deviations in the prediction of a single wind farm due to the interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a system block diagram of the present invention.

[0052] Figure 2 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0053] Wind power refers to the rate at which wind turbines in a wind farm convert wind energy into electrical energy. The prediction of wind power is crucial to the operation and management of wind farms because it affects the power generation of wind farms and the dispatch of power grids. Seasonal and climate changes and wind speed and direction will affect the operating conditions of wind farms. In order to maintain wind power, it is necessary to predict the power of wind farms. Existing prediction systems usually rely on historical data and numerical weather forecasts, which may be missing or inaccurate, affecting the prediction accuracy. Some prediction models may perform well under specific conditions, but lack generalization capabilities under other conditions, resulting in large prediction errors. Based on this, an embodiment of the present invention provides a wind power prediction system based on solar terms, which is as follows:

[0054] A solar term-based wind power prediction system includes a data acquisition module, a data calibration module, a real-time prediction module and a cluster prediction module, wherein the data acquisition module is used to collect historical data and environmental data of the current environment; it should be noted that the environmental data includes wind speed, air density, air pressure, temperature and humidity, which are obtained through various sensors;

[0055] The data calibration module is used to calibrate the information obtained by the data acquisition module to reduce the error of the data; it should be noted that the data calibration module is used to calibrate the data obtained by the sensor according to historical data to reduce the error caused by the measurement;

[0056] The real-time prediction module is used to perform power prediction on the wind farm according to the data calibrated by the data calibration module in a combined manner, specifically:

[0057] The physical method predicted power value of the wind farm is obtained and marked as Q; it should be noted that the physical method is used to predict the predicted power of the wind farm based on the wind speed and the environmental information of the wind farm;

[0058] The power value predicted by the time series method of the wind farm is obtained and marked as W; it should be noted that the power value predicted by the time series method is to predict the predicted power of the wind farm by using the time series analysis method;

[0059] The power value predicted by the artificial intelligence method of the wind farm is obtained and marked as E; it should be noted that the power value predicted by the artificial intelligence method is to predict the wind power by using a neural network;

[0060] According to the formula The final predicted power value R is calculated and obtained, where , and is the preset proportional coefficient; it should be noted that , and It is set by the staff according to experience. In this embodiment, , and It can be 0.428, 0.376 and 0.196;

[0061] A threshold value of the predicted power value is set in advance and marked as T. The predicted power value R is compared with the threshold value T. If the predicted power value R is less than the threshold value T, it is judged that the power generation power of the wind farm is insufficient. If not, it is judged that the wind farm is operating normally.

[0062] It should be noted that the predicted powers obtained by different methods are combined and calculated to obtain the final predicted power value, so as to avoid the situation where some prediction models may perform well under certain conditions but have insufficient generalization ability under other conditions, resulting in large prediction errors;

[0063] The cluster prediction module is used to comprehensively consider the interaction between multiple wind farms and the overall meteorological conditions to further predict the predicted power of the entire wind farm cluster.

[0064] As an optional embodiment, the physical method predicted power value of the wind farm is obtained as follows:

[0065] The efficiency value of the wind turbine generator set in the wind farm is obtained and marked as It should be noted that the efficiency value is an integer from 1 to 20, obtained through staff evaluation, and is used to quantify the working efficiency of the wind turbine;

[0066] According to the formula , calculating and obtaining the physical method predicted power value of the wind farm;

[0067] in is the air density value, d is the radius of the wind wheel in the wind power equipment, and V is the ambient wind speed.

[0068] As an optional embodiment, the method for obtaining the power value predicted by the time series method of the wind farm is as follows, specifically:

[0069] Get the predicted power value one year before the current time point and mark it as ;

[0070] According to the formula , calculate and obtain the time series method predicted power value of the wind farm ;

[0071] Wherein, b is a preset smoothing parameter. In this embodiment, the value of b may be 1.293.

[0072] As an optional embodiment, the method for obtaining the power value predicted by the artificial intelligence method of the wind farm is as follows, specifically:

[0073] According to the formula , calculate and obtain the artificial intelligence predicted power value E of the wind farm;

[0074] in is the activation function, is the input feature vector, which is any one of wind speed, air density, air pressure, temperature and humidity. is the weight corresponding to each feature vector, n is the number of input features, which is 5, and b is the bias term. It should be noted that the activation function is the implementation of nonlinear transformation in the neural network. They introduce nonlinearity so that the neural network can learn and perform more complex tasks. In this embodiment, the activation function can be , that is, when When X is less than zero, output 0, and when X is greater than zero, output X;

[0075] The bias term is a constant added to the weighted sum of the neural network inputs. Its function is to provide the model with translational freedom so that the decision boundary can better fit the data. In this embodiment, the bias term is 4.12;

[0076] The weight value corresponding to each eigenvector is changed according to wind speed, air density, air pressure, temperature and humidity. In this embodiment, W is the weight value of wind speed, air density, air pressure, temperature and humidity, which are 0.987, 0.928, 0.978, 0.897 and 0.596 respectively;

[0077] As an optional embodiment, the specific working mode of the data calibration module is as follows:

[0078] Get the real-time environmental information collected by the data acquisition module and mark it as U;

[0079] The environmental information is any one of wind speed, air density, air pressure, temperature and humidity;

[0080] According to the formula , calculate and obtain the calibrated environmental information Y, where The default value of the environment information, It is a preset correction factor. It should be noted that calibration can reduce the possibility of missing or inaccurate data, which may affect the prediction accuracy.

[0081] As an optional embodiment, the preset value of the environmental information The method of obtaining is as follows:

[0082] Get the node dates of the solar terms in the five years before the current time period, divide the environmental information obtained according to the time nodes of the 24 solar terms, establish 24 groups of data sample sets corresponding to the solar terms, get the average of the environmental information data recorded in the data sample set, and mark it as the preset value .

[0083] As an optional embodiment, the preset correction coefficient The way to obtain is as follows:

[0084] Get each detected environment information and preset value in the year before the current time period Deviation;

[0085] Get the average of all deviations and mark it as P;

[0086] The standard deviation of all deviations is calculated and marked as ;

[0087] According to the formula , calculate and obtain the preset correction coefficient .

[0088] It should be noted that existing wind farms are installed in clusters. The power prediction of wind farm clusters needs to consider the interaction of multiple wind farms and the influence of overall meteorological conditions. The power prediction of a single wind farm may not be very accurate. This technical solution can solve the above problems, as follows;

[0089] As an optional embodiment, the specific working mode of the cluster prediction module is as follows:

[0090] Taking the wind farm as the center and the preset radius as the center, a circle is drawn to obtain the cluster area;

[0091] Obtain the difference between the predicted power value R and the threshold value T of all wind farms in the cluster area;

[0092] A threshold of a difference is set in advance, all differences are compared with the threshold, all wind farms with differences less than the threshold are marked as benchmark wind farms, and the number of benchmark wind farms is marked.

[0093] As an optional embodiment, the total installed capacity of the wind farms in the cluster area is obtained and marked as H;

[0094] According to the formula , calculate and obtain the total predicted power L of the cluster area, where is the power forecast value of the i-th benchmark wind farm, is the installed capacity of the i-th benchmark wind farm;

[0095] It should be noted that by considering the interaction of multiple wind farms, the total predicted power of the entire wind farm cluster can be obtained, thereby avoiding large deviations in the prediction of a single wind farm due to the interaction.

[0096] like Figure 2 As shown: The present invention also proposes a wind power prediction method based on solar terms, comprising the following steps:

[0097] Step 1: Collect historical data and numerical weather forecasts, and calibrate the data to reduce data errors;

[0098] Step 2: Use the calibrated data to predict the power of the wind farm and determine whether the power of the wind farm meets expectations;

[0099] Step 3: Based on the obtained predicted power, the predicted power of the entire wind farm cluster is further predicted by combining the interaction of multiple wind farms and the overall meteorological conditions.

[0100] Working principle: By setting up a data calibration module and a real-time prediction module, the data obtained by the sensor can be calibrated according to historical data to reduce the error caused by the measurement. The predicted power obtained by different methods is combined and calculated to obtain the final predicted power value, avoiding the situation where some prediction models may perform well under certain conditions, but lack generalization ability under other conditions, resulting in large prediction errors;

[0101] By providing a cluster prediction module, the interaction between multiple wind farms can be considered to obtain the total predicted power of the entire wind farm cluster, thereby avoiding large deviations in the prediction of a single wind farm due to the interaction.

[0102] The above are only preferred implementations of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical staff in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of this template.

Claims

1. A wind power prediction system based on solar terms, characterized in that: include: A data collection module, wherein the data collection module is used to collect historical data and environmental data of the current environment; A data calibration module, the data calibration module is used to calibrate the information acquired by the data acquisition module; A real-time prediction module is used to perform power prediction on a wind farm according to the data calibrated by the data calibration module in a combined manner, specifically: Obtain the physical predicted power value of the wind farm and mark it as Q; Obtain the wind farm's time series predicted power value and mark it as W; Obtain the power value predicted by the artificial intelligence method of the wind farm and mark it as E; According to the formula The final predicted power value R is calculated and obtained, where , and is the preset proportional coefficient; A threshold value of the predicted power value is set in advance and marked as T. The predicted power value R is compared with the threshold value T. If the predicted power value R is less than the threshold value T, it is judged that the power generation power of the wind farm is insufficient. If not, it is judged that the wind farm is operating normally. A cluster prediction module, which is used to further predict the predicted power of the entire wind farm cluster by integrating the interaction of multiple wind farms and overall meteorological conditions; The specific working mode of the data calibration module is as follows: Get the real-time environmental information collected by the data acquisition module and mark it as U; The environmental information is any one of wind speed, air density, air pressure, temperature and humidity; According to the formula , calculate and obtain the calibrated environmental information Y, where The default value of the environment information, is the preset correction factor; The preset value of the environmental information The method of obtaining is as follows: Get the node dates of the solar terms in the five years before the current time period, divide the environmental information obtained according to the time nodes of the 24 solar terms, establish 24 groups of data sample sets corresponding to the solar terms, get the average of the environmental information data recorded in the data sample set, and mark it as the preset value ; The preset correction factor The way to obtain is as follows: Get each detected environment information and preset value in the year before the current time period Deviation; Get the average of all deviations and mark it as P; The standard deviation of all deviations is calculated and marked as ; According to the formula , calculate and obtain the preset correction coefficient .

2. The wind power prediction system based on solar terms according to claim 1, characterized in that: The method for obtaining the power value predicted by the physical method of the wind farm is as follows: The efficiency value of the wind turbine generator set in the wind farm is obtained and marked as According to the formula , calculating and obtaining the physical method predicted power value of the wind farm; in is the air density value, d is the radius of the wind wheel in the wind turbine, and V is the wind speed.

3. The wind power prediction system based on solar terms according to claim 1, characterized in that: The method for obtaining the power value predicted by the time series method of the wind farm is as follows: Get the predicted power value one year before the current time point and mark it as According to the formula , calculate and obtain the time series method predicted power value of the wind farm ; Among them, b is the preset smoothing parameter.

4. The wind power prediction system based on solar terms according to claim 1, characterized in that: The method for obtaining the power value predicted by the artificial intelligence method of the wind farm is as follows, specifically: According to the formula , calculate and obtain the artificial intelligence predicted power value E of the wind farm; in is the activation function, is the input feature vector, which is any one of wind speed, air density, air pressure, temperature and humidity. is the weight corresponding to each feature vector, n is the number of input features, which is 5, and b is the bias term.

5. The wind power prediction system based on solar terms according to claim 1, characterized in that: The specific working mode of the cluster prediction module is as follows: Taking the wind farm as the center and the preset radius as the center, a circle is drawn to obtain the cluster area; Obtain the difference between the predicted power value R and the threshold value T of all wind farms in the cluster area; A difference threshold is set in advance, all differences are compared with the threshold, all wind farms with differences less than the threshold are marked as benchmark wind farms, and the number of benchmark wind farms is marked as K.

6. The wind power prediction system based on solar terms according to claim 5, characterized in that: The total installed capacity of the wind farms in the cluster area is obtained and marked as H; According to the formula , calculate and obtain the total predicted power L of the cluster area, where is the power forecast value of the i-th benchmark wind farm, is the installed capacity of the ith benchmark wind farm.

7. A method for predicting wind power based on solar terms, applicable to the wind power prediction system based on solar terms as claimed in claims 1 to 6, characterized in that: The following steps are involved: Step 1: Collect historical data and numerical weather forecasts, and calibrate the data to reduce data errors; Step 2: Use the calibrated data to predict the power of the wind farm and determine whether the power of the wind farm meets expectations; Step 3: Based on the obtained predicted power, the predicted power of the entire wind farm cluster is further predicted by integrating the interaction of multiple wind farms and the overall meteorological conditions; The specific working method of step 1 includes: obtaining real-time environmental information collected by the data acquisition module; The environmental information is any one of wind speed, air density, air pressure, temperature and humidity; Obtain the node dates of the solar terms in the five years before the current time period, divide the environmental information obtained according to the time nodes of the 24 solar terms, establish 24 groups of data sample sets corresponding to the solar terms, obtain the average of the environmental information data recorded in the data sample set, mark it as the preset value, and calculate and obtain the calibrated environmental information Y according to the preset value.

Images