Short-term wind power forecasting method based on progressive deep learning of multi-source data

Abstract

Disclosed in the present invention is a short-term wind power forecasting method based on progressive deep learning of multi-source data, comprising the following steps: S1, training a forecasting model by means of a deep learning network structure; and S2, computing wind speed data of each wind turbine by means of the forecasting model, and carrying out wind power forecasting. The present invention has high forecasting accuracy; a wind speed forecasting model is trained on the basis of multi-source data, and thus, the forecasting accuracy of computing the actual wind speed can be greatly improved.

Description

A short-term wind power forecasting method based on progressive deep learning of multi-source data Technical Field

[0001] The present invention relates to the technical field of wind power prediction, and in particular to a short-term wind power prediction method based on progressive deep learning of multi-source data. Background Art

[0002] Wind power forecasting is crucial in the renewable energy sector. With the increasing global demand for clean energy, wind power, as a green and sustainable energy source, is gaining increasing attention. However, the instability and randomness of wind power pose challenges to power system operation. Therefore, accurate wind power forecasting, especially short-term wind power forecasting, is crucial for reducing wind curtailment, optimizing daily power generation plans and cold and hot standby systems for conventional power sources, and adjusting maintenance plans to more accurately reflect the utilization efficiency of wind energy resources. This helps optimize power system operation and ensure the stability and economic efficiency of power supply.

[0003] Wind power prediction, its current relevant data often comes from weather forecasts, which make certain simplifications and approximations to the basic equations of atmospheric motion, and introduce meteorological physical laws and boundary condition constraints. The time resolution of wind speed forecasts in numerical weather forecasts is generally 1 hour or 3 hours, and the spatial resolution is generally 9km×9km or 12.5km×12.5km. For wind farms, the generally required time resolution of wind power prediction is 15 minutes, and wind speed forecasts at the wind turbine locations are required. Therefore, the current relevant wind power prediction methods cannot fully meet actual needs. Among the current related technologies, the main method is to predict wind power for a large area containing a wind farm. It is still difficult to predict the wind power of each wind turbine in the wind farm. The wind power prediction data for a large area is not applicable to each specific wind turbine therein.

[0004] Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a short-term wind power prediction method based on progressive deep learning of multi-source data to solve the problems raised in the background technology.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is: a short-term wind power prediction method based on progressive deep learning of multi-source data, which includes the following steps:

[0007] S1: Use deep learning network structure to train prediction model;

[0008] S2: Use the prediction model to calculate the wind speed data of each wind turbine and perform wind power prediction;

[0009] Furthermore, the step S1 specifically includes the following process:

[0010] S1.1. Collect multi-source data required for the prediction model;

[0011] S1.2, preprocess the multi-source data collected in step 1.1;

[0012] S1.3, divide the processed data set;

[0013] S1.4. Train the deep learning model to obtain the prediction model function f(x).

[0014] Furthermore, the step S2 specifically includes the following process:

[0015] S2.1. Collect weather forecast data for a certain wind farm;

[0016] S2.2. Use the trained prediction model f(x) to calculate the wind speed of a wind turbine over a period of time.

[0017] S2.3. Obtain wind speed data and wind power data of a certain wind turbine after a period of time.

[0018] Furthermore, step S1.1 specifically includes: collecting numerical values ​​related to wind speed of each wind turbine and weather forecast data to form a multi-source data set, wherein the data related to wind speed includes two parts: one is weather forecast data updated for a period of time for the wind farm, including wind direction, short-term weather type, rainfall type, and temperature; the other is data to be collected at each wind turbine location, including wind turbine height, terrain type, temperature, air pressure, humidity, sea level pressure, downward shortwave radiation from the ground, downward longwave radiation from the ground, and total cloud cover;

[0019] The multi-source data are labeled and divided into two groups. One group is the real-time collected data X1, X2...X at each wind turbine location. n The other group is the weather forecast data Y1, Y2...Y for each wind turbine in real time. n .

[0020] Furthermore, in step S1.2, specific measures for data preprocessing include: correcting or deleting incomplete or erroneous data, processing duplicate data, processing outliers, unifying data formats, data normalization and data encoding.

[0021] Furthermore, the step S1.3 is specifically as follows: dividing the preprocessed multi-source data into a training set, a validation set and a test set according to a certain ratio, wherein the training set is a data set used to train the machine learning model, the validation set is a data set set aside separately for evaluating model performance and adjusting hyperparameters during the deep learning model training process, and the test set is a data set used to finally evaluate the performance of the machine learning model.

[0022] Furthermore, in step S1.4, the deep learning model includes a deep learning neural network structure, and the deep neural network includes an input layer, an output layer, and at least one hidden layer; the network parameters are adjusted, and the number of network layers, the number of neurons, batch normalization, and random dropout are adjusted to train the network until the maximum number of iterations is reached or the network learning rate converges; mean square error is selected as the loss function and mean absolute error is selected as the evaluation accuracy; the number of network layers and the number of neurons are set and adjusted according to the application in different regional scenarios until the loss function converges or the maximum number of iterations is reached, thereby completing the model training;

[0023] Collect multi-source data X1, X2...X at each wind turbine location in step S1.1 in real time n Collect data from each wind turbine in real time, including weather forecast data from multiple sources Y1, Y2...Y n As the input layer, the number of hidden layers is set according to the amount of data, at least more than 1 layer. After adjusting the activation function, weight and other parameters, the real-time multi-source data X1, X2...X of each wind turbine is obtained through training. n Compared with the multi-source data Y1, Y2...Y provided by the weather forecast in the previous period n The previous relationship is to obtain the prediction model function f(x).

[0024] Furthermore, the wind farm weather forecast data collected in step S2.1 includes weather forecast data Y1, Y2...Y for a large area of ​​a certain wind farm. n ;

[0025] The step S2.2 is specifically as follows: using the trained wind speed prediction model f(x) for each wind turbine, combined with the weather forecast data Y1, Y2...Y collected in step S2.1 n , calculate the wind speed data at each wind turbine location after a period of time;

[0026] The step S2.3 is specifically: obtaining the wind speed data at each wind turbine position after a period of time, and then accurately predicting the wind power data of a specific wind turbine after a period of time.

[0027] Beneficial effects of the present invention:

[0028] 1. The present invention has high prediction accuracy. It uses a wind speed prediction model based on multi-source data, which can greatly improve the prediction accuracy of the actual wind speed calculation;

[0029] 2. The present invention conducts progressive deep learning on multi-source data to train the prediction model. Progressive deep learning refers to deep learning of data such as the current actual wind speed and data provided by the weather forecast for a period of time. That is, deep learning is conducted progressively over a period of time. It can learn the historical parameters related to the wind speed of each wind turbine. It can overcome the limitations of existing forecast accuracy, which is restricted by existing meteorological scientific theories, and the time resolution and spatial resolution of numerical weather forecasts.

[0030] 3. The present invention breaks through the limitation of being unable to accurately predict the wind speed of each specific wind turbine. After long-term progressive deep learning of the wind turbines at each specific location, the prediction model function after progressive deep learning is obtained. Then, based on traditional weather forecast data, the wind speed changes of each wind turbine after a period of time can be accurately predicted. This breaks through the problem that the original weather forecast data cannot be fully applied to the actual situation of a specific wind turbine for large-area forecasts containing wind farms. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] FIG1 is a schematic diagram of a process for training a prediction model using a deep learning network structure according to the present invention;

[0032] FIG2 is a schematic diagram of a process for calculating the wind speed data of each wind turbine using a prediction model according to the present invention;

[0033] Figure 3 is a schematic diagram of the logical structure of the prediction model trained using a deep learning network structure in the present invention. DETAILED DESCRIPTION

[0034] The present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0035] Example 1: As shown in Figures 1 to 3, a short-term wind power prediction method based on progressive deep learning of multi-source data includes the following steps:

[0036] S1: Use deep learning network structure to train prediction model;

[0037] S1.1. Collect multi-source data required for the prediction model;

[0038] S1.2, preprocess the multi-source data collected in step 1.1;

[0039] S1.3, divide the processed data set;

[0040] S1.4. Train the deep learning model to obtain the prediction model function f(x).

[0041] S2: Use the prediction model to calculate the wind speed data of each wind turbine and perform wind power prediction;

[0042] S2.1. Collect weather forecast data for a certain wind farm;

[0043] S2.2. Use the trained prediction model f(x) to calculate the wind speed of a wind turbine over a period of time.

[0044] S2.3. Obtain wind speed data and wind power data of a certain wind turbine after a period of time.

[0045] Furthermore, step S1.1 specifically includes: collecting numerical values ​​related to wind speed of each wind turbine and weather forecast data to form a multi-source data set, wherein the data related to wind speed includes two parts: one is weather forecast data updated for a period of time for the wind farm, including wind direction, short-term weather type, rainfall type, and temperature; the other is data to be collected at each wind turbine location, including wind turbine height, terrain type, temperature, air pressure, humidity, sea level pressure, downward shortwave radiation from the ground, downward longwave radiation from the ground, and total cloud cover;

[0046] The multi-source data are labeled and divided into two groups. One group is the real-time collected data X1, X2...X at each wind turbine location. n The other group is the weather forecast data Y1, Y2...Y for each wind turbine in real time. n .

[0047] Furthermore, in step S1.2, specific measures for data preprocessing include: correcting or deleting incomplete or erroneous data, processing duplicate data, processing outliers, unifying data formats, data normalization and data encoding.

[0048] Furthermore, the step S1.3 is specifically as follows: dividing the preprocessed multi-source data into a training set, a validation set and a test set according to a certain ratio, wherein the training set is a data set used to train the machine learning model, the validation set is a data set set aside separately for evaluating model performance and adjusting hyperparameters during the deep learning model training process, and the test set is a data set used to finally evaluate the performance of the machine learning model.

[0049] Furthermore, in step S1.4, the deep learning model includes a deep learning neural network structure, and the deep neural network includes an input layer, an output layer, and at least one hidden layer; the network parameters are adjusted, and the number of network layers, the number of neurons, batch normalization, and random dropout are adjusted to train the network until the maximum number of iterations is reached or the network learning rate converges; mean square error is selected as the loss function and mean absolute error is selected as the evaluation accuracy; the number of network layers and the number of neurons are set and adjusted according to the application in different regional scenarios until the loss function converges or the maximum number of iterations is reached, thereby completing the model training;

[0050] Collect multi-source data X1, X2...X at each wind turbine location in step S1.1 in real time n Collect data from each wind turbine in real time, including weather forecast data from multiple sources Y1, Y2...Y n As the input layer, the number of hidden layers is set according to the amount of data, at least more than 1 layer. After adjusting the activation function, weight and other parameters, the real-time multi-source data X1, X2...X of each wind turbine is obtained through training. n Compared with the multi-source data Y1, Y2...Y provided by the weather forecast in the previous period n The previous relationship is to obtain the prediction model function f(x).

[0051] Furthermore, the wind farm weather forecast data collected in step S2.1 includes weather forecast data Y1, Y2...Y for a large area of ​​a certain wind farm. n ;

[0052] The step S2.2 is specifically as follows: using the trained wind speed prediction model f(x) for each wind turbine, combined with the weather forecast data Y1, Y2...Y collected in step S2.1 n , calculate the wind speed data at each wind turbine location after a period of time;

[0053] The step S2.3 is specifically: obtaining the wind speed data at each wind turbine position after a period of time, and then accurately predicting the wind power data of a specific wind turbine after a period of time.

[0054] Example 2:

[0055] S1: Use deep learning network structure to train prediction model;

[0056] The specific steps are detailed as follows:

[0057] Step 1.1: Collect the multi-source data required for the prediction model. Collect wind speed-related numerical values ​​and weather forecast data for each wind turbine to form a multi-source dataset. Wind speed-related data consists of two parts: one is the weather forecast data updated every hour for the wind farm, such as wind direction, short-term weather type, rainfall type, and temperature; the other is the data required to be collected at each wind turbine location, such as turbine height, terrain type, temperature, air pressure, humidity, sea level pressure, downward shortwave radiation from the ground, downward longwave radiation from the ground, and total cloud cover.

[0058] Note: The "1h" mentioned here and in this patent is an example for the convenience of explanation and does not refer to only 1h. It represents the time interval for updating the current weather forecast data or the time interval set by each wind farm that is greater than or equal to this time interval. It can be 1.5h, 2h, etc., which are all within the scope of protection and description of this patent.

[0059] The multi-source data from step 1.1 can be labeled as shown in Tables 1 and 2 below. Table 1 shows the real-time data collected at each wind turbine location, and Table 2 shows the weather forecast data for the hour before the real-time data was collected for each wind turbine. For example, if the entire day is divided by hour, there are 24 sets of data. For example, at 12:00 (24-hour system), Table 1 shows the real-time multi-source data collected at each wind turbine location, and Table 2 shows the data provided by the weather forecast for the entire wind farm area at 11:00.

[0060] Table 1 Multi-source data labels collected in real time at each wind turbine location (12 o'clock)

[0061] Table 2 Weather forecast multi-source data labels for each wind turbine 1 hour before real-time data collection (11:00)

[0062] Step 1.2: Preprocess the multi-source data collected in Step 1.1. Data preprocessing aims to improve data quality, consistency, and completeness, making it more suitable for analysis or mining. This series of operations includes processing "dirty" data, accurately extracting data, and adjusting data formats, thereby obtaining high-quality data that meets standards for accuracy, completeness, and conciseness. Specific measures for data preprocessing include correcting or deleting incomplete or erroneous data, handling duplicate data, addressing outliers, standardizing data formats, normalizing data, and encoding data.

[0063] ① Correcting or deleting incomplete or erroneous data, processing duplicate data, and handling outliers can be divided into the following steps:

[0064] 1) Identify incomplete or erroneous data: This can be achieved by checking the data for outliers, missing values, inconsistent data, or other obvious errors. 2) Fill in missing values: This can be handled using simple deletion or weighting methods. Simple deletion directly deletes cases with missing values. Weighting methods weight complete data to reduce bias, and case weights can be obtained through logistic or probit regression. 3) Deal with outliers: Outliers (abnormal values) are the norm in data distribution. Data outside a specific distribution area or range is usually defined as anomalies or noise, and can usually be handled by deleting outliers.

[0065] ② Unifying data formats, data normalization, and data encoding can be divided into the following steps:

[0066] 1) Unify data formats: If the data formats are not unified, you can use corresponding methods to convert them. You can use Python's pandas library to convert the data format; 2) The normalization formula can be used as follows; 3) Data encoding can use encoding methods such as one-hot encoding and label encoding.

[0067] Step 1.3: Divide the processed dataset. Divide the preprocessed multi-source data into a training set, a validation set, and a test set in a ratio of 7:1.5:1.5. The training set is the dataset used to train the machine learning model. The validation set is a dataset set aside during the deep learning model training process for evaluating model performance and adjusting hyperparameters. The test set is the dataset used to ultimately evaluate the performance of the machine learning model.

[0068] Step 1.4: Train the deep learning model. The deep learning model includes various deep learning neural network structures. The deep neural network should include an input layer, an output layer, and at least one hidden layer. The network parameters can be adjusted by adjusting the number of network layers, the number of neurons, batch normalization, and random inactivation to train the network until the maximum number of iterations is reached or the network learning rate converges. At the same time, the mean square error is selected as the loss function and the mean absolute error is selected as the evaluation accuracy. Of course, the number of network layers and the number of neurons here can be set and adjusted according to the application in different regional scenarios until the value of the loss function converges or the maximum number of iterations is reached, and the model training is completed.

[0069] The real-time multi-source data X1, X2...X at each wind turbine location in Table 1 in step 1.1 are collected n Compared with the weather forecast multi-source data Y1, Y2...Y nAs the input layer, the number of hidden layers can be set according to different data amounts and should be at least greater than 1 layer. By adjusting parameters such as activation functions and weights, the relationship between the real-time multi-source data of each wind turbine and the multi-source data provided by the weather forecast for the previous hour can be trained to obtain the prediction model function f(x).

[0070] S2: Use the prediction model to calculate the wind speed data of each wind turbine and perform wind power prediction;

[0071] The specific steps are detailed as follows:

[0072] Step 2.1: Collect weather forecast data Y1, Y2, ...Y for a large area containing a wind farm at a certain time. n wait;

[0073] Step 2.2: Use the trained wind speed prediction model f(x) for each wind turbine and combine it with the weather forecast data Y1, Y2...Y collected in step 2.1. n , calculate the wind speed data at each fan location after 1 hour;

[0074] Step 2.3: Calculate the wind speed data at each wind turbine location 1 hour later, and then accurately predict the specific wind power data of a specific wind turbine 1 hour later.

[0075] For Example 2, the following supplementary explanations are given:

[0076] 1. The "1h" expressed in this patent is an example for the convenience of explanation and does not refer to only 1h. It represents the time interval for updating the current weather forecast data or the time interval set by each wind farm that is greater than or equal to this time interval. It can be 1.5h, 2h, etc., all of which are within the scope of protection of this patent.

[0077] 2. In the prediction model training phase, the multi-source data in step 1 is only partially listed in this patent, including wind turbine height, terrain type, temperature, air pressure, humidity, sea level pressure, ground-down shortwave radiation, ground-down longwave radiation, total cloud cover, etc. This list is not exhaustive, and some parameters may be deleted or supplemented based on actual conditions, and all of these parameters shall be within the scope of protection of this patent.

[0078] 3. In the prediction model training stage, the deep learning network structure in step 4 does not specifically refer to a specific deep learning algorithm, such as a convolutional neural network, and all deep learning neural network structures should be within the scope of protection of this patent.

[0079] The above embodiments are merely preferred technical solutions of the present invention and should not be construed as limiting the present invention. The scope of protection of the present invention shall be the technical solutions set forth in the claims, including equivalent alternatives to the technical features of the technical solutions set forth in the claims. In other words, equivalent alternatives and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A short-term wind power prediction method based on multi-source data progressive deep learning, characterized by: It includes the following steps: S1: Use deep learning network structure to train prediction model; S2: Use the prediction model to calculate the wind speed data of each wind turbine and predict the wind power; 2. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 1 is characterized in that: The step S1 specifically includes the following process: S1.

1. Collect multi-source data required for the prediction model; S1.2, preprocessing the multi-source data collected in step 1.1; S1.3, dividing the processed data set; S1.

4. Train the deep learning model to obtain the prediction model function f(x).

3. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 2 is characterized in that: The step S2 specifically includes the following process: S2.

1. Collect weather forecast data for a certain wind farm; S2.2, using the trained prediction model f(x), calculate the wind speed of a certain wind turbine after a period of time; S2.

3. Obtain the wind speed data and wind power data of a certain wind turbine after a period of time.

4. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 2 is characterized in that: The step S1.1 is specifically as follows: collecting the numerical values ​​and weather forecast data related to wind speed of each wind turbine to form a multi-source data set, wherein the data related to wind speed includes two parts: one is the weather forecast data updated for a period of time for the wind farm, including wind direction, short-term weather type, rainfall type, and temperature; the other is the data to be collected at each wind turbine location, including wind turbine height, terrain type, temperature, air pressure, humidity, sea level pressure, downward short-wave radiation from the ground, downward long-wave radiation from the ground, and total cloud cover; The multi-source data are labeled and divided into two groups. One group is the real-time collected data X1, X2...X at each wind turbine location. n The other group is the weather forecast data Y1, Y2...Y for each wind turbine in real time. n .

5. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 2 is characterized in that: In step S1.2, specific measures for data preprocessing include: correcting or deleting incomplete or erroneous data, processing duplicate data, processing outliers, unifying data formats, data normalization and data encoding.

6. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 2 is characterized in that; The step S1.3 is specifically as follows: the preprocessed multi-source data is divided into a training set, a validation set and a test set according to a certain ratio, wherein the training set is a data set used to train the machine learning model, the validation set is a data set set aside separately during the deep learning model training process for evaluating model performance and adjusting hyperparameters, and the test set is a data set used to finally evaluate the performance of the machine learning model.

7. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 2 is characterized by: In the step S1.4, the deep learning model includes various deep learning neural network structures, and the deep neural network includes an input layer, an output layer and at least one hidden layer; the network parameters are adjusted, and the number of network layers, the number of neurons, batch normalization, and random inactivation are adjusted to train the network until the maximum number of iterations is reached or the network learning rate converges; mean square error is selected as the loss function and mean absolute error is selected as the evaluation accuracy; the number of network layers and the number of neurons are set and adjusted according to the application in different regional scenarios until the value of the loss function converges or the maximum number of iterations is reached, and the model training is completed; Collect multi-source data X1, X2...X n Collect data from each wind turbine in real time, and the weather forecast multi-source data Y1, Y2...Y n As the input layer, the number of hidden layers is set according to the amount of data, at least more than 1 layer. By adjusting the activation function, weight and other parameters, the real-time multi-source data X1, X2...X of each wind turbine is obtained after training. n Compared with the multi-source data Y1, Y2...Y provided by the weather forecast in the previous period n The previous relationship gives the prediction model function f(x).

8. The short-term wind power prediction method based on multi-source data progressive deep learning according to claim 3 is characterized by: The wind farm weather forecast data collected in step S2.1 includes weather forecast data Y1, Y2, ..., Y n ; The step S2.2 is specifically as follows: using the trained wind speed prediction model f(x) for each wind turbine, combined with the weather forecast data Y1, Y2...Y n , calculate the wind speed data at each wind turbine location after a period of time; The step S2.3 is specifically: obtaining the wind speed data at each wind turbine position after a period of time, and then accurately predicting the wind power data of a specific wind turbine after a period of time.

Images