Method and device for predicting power curve of wind generating set

Abstract

The invention provides a wind generating set power curve prediction method and device, and relates to the technical field of wind generating sets, and the method comprises the steps: firstly collecting the actual power of a wind generating set, the rotating speed of a generator and environment data, and providing a basis for subsequent analysis; then, the power generation data at the previous moment serve as input, the actual power at the next moment serves as a label, a multiple linear regression model is trained, the influence of the current moment is focused, the previous accumulated influence is easily ignored, then the power generation data at the previous moments serve as input, and the actual power at the next moment serves as the label; the training of the long and short-term memory network model focuses on the influence of time series data and is suitable for dynamically changing data. And the rotating speed, wind speed, temperature and humidity data of the generator are analyzed, a fluctuation adjustment coefficient is obtained, the fluctuation adjustment coefficient is applied to correct the actual power predicted by the two models at the current moment, and the corrected actual power is applied to the power curve of the wind generating set.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbines, and specifically to a method and device for predicting the power curve of a wind turbine. Background Art

[0002] Wind power generation, as an important renewable energy source, has become increasingly prominent in the global energy structure in recent years. With the growing global concern for sustainable development, the use of traditional fossil fuels faces increasing environmental pressure, leading to problems such as greenhouse gas emissions and air pollution. Due to its clean and renewable characteristics, wind energy has become one of the alternative energy sources actively promoted by governments of various countries, contributing to reducing carbon emissions, protecting the environment, and enhancing energy security.

[0003] The prediction of the power curve of a wind turbine is a crucial link in the utilization and management of wind energy. The power curve describes the power generation capacity of a wind turbine at different wind speeds, and accurate prediction can effectively improve the efficiency of wind power generation and the stability of the power grid. The following is the background art related to the prediction of the power curve of a wind turbine. First of all, the establishment of the power curve is usually based on the relationship between wind speed and output power. The power output of a wind turbine is affected by wind speed and usually shows non-linear characteristics. Traditional power curve prediction methods mostly rely on the statistical analysis of historical data, including regression analysis and time series models, etc. However, these methods often cannot provide sufficient accuracy when facing complex and dynamic wind speed changes. Therefore, in recent years, more and more research has started to adopt machine learning and data-driven methods to improve prediction accuracy. These methods analyze a large amount of historical data, extract features and patterns from it, enabling the prediction model to better adapt to wind speed changes and improve the generalization ability of the model.

[0004] The existing technologies have the following deficiencies:

[0005] There are multiple deficiencies in the existing power curve prediction technologies for wind turbines, mainly including the neglect of the non - linear relationship between wind speed and actual power, poor adaptability to dynamic changes, insufficient model generalization ability, and lack of real - time performance. These can all lead to a reduction in prediction accuracy and reliability. Many traditional prediction methods mainly use linear regression models, assuming that the relationship between wind speed and power generation is linear. However, in fact, this relationship is often complex and non - linear. Especially when the wind speed changes greatly, the linear model is prone to increased prediction errors and cannot accurately reflect the actual power generation situation. In addition, traditional methods have poor adaptability in dealing with rapidly changing environmental conditions, making it difficult for them to capture in a timely manner the impact of changes in factors such as wind speed and temperature on power generation, resulting in lagged prediction results. Moreover, many models perform well on specific data sets, but in different wind farms or different geographical conditions, the model generalization ability is insufficient and cannot be widely applied to various situations, limiting the scope of its practical application. Some traditional prediction methods lack real - time performance and cannot be dynamically updated according to the latest data, which makes the model unable to reflect the actual state of the current wind turbine and affects the timeliness and effectiveness of scheduling decisions.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a method and device for predicting the power curve of a wind turbine to solve the problems raised in the above - mentioned background art.

[0008] To achieve the above - mentioned purpose, the present invention provides the following technical solutions:

[0009] A method for predicting the power curve of a wind turbine, the specific steps include:

[0010] Step 1: Collect historical power generation data of the wind power generation unit. The historical power generation data includes the actual power, generator speed, and environmental data at each moment in the power curve of the generator set. The environmental data includes wind speed, temperature, and humidity;

[0011] Step 2: Use the actual power, generator speed, and environmental data at the previous moment in the historical power generation data as inputs, and the actual power at the next moment as a label to construct and train a multiple linear regression model;

[0012] Step 3: Use the historical power generation data of the previous m moments in the historical power generation data as inputs, and the actual power at the next moment as a label to construct and train a power prediction model based on the long - short - term memory network model;

[0013] Step 4: Calculate the fluctuation adjustment coefficient according to the generator speed, wind speed, temperature and humidity data at the current moment.

[0014] Step 5: Collect the power generation data at the current moment and the previous m - 1 moments, input them into the multiple linear regression model and the power prediction model, obtain the actual power at the next moment predicted by the two models, and then calculate the actual power corrected by the fluctuation adjustment coefficient in combination with the fluctuation adjustment coefficient, and apply the corrected actual power to the power curve of the wind turbine generator set.

[0015] Further, the specific process of collecting the historical power generation data of the wind turbine generator set includes:

[0016] Collect the current, voltage and generator speed data of the wind turbine generator set through the power sensor and speed sensor equipped on the wind turbine generator set, obtain the power factor of the wind turbine generator set through the computing type watt-hour meter, and the formula for calculating the actual power of the wind turbine generator set is expressed as:

[0017]

[0018] where P is the actual power, V is the voltage, I is the current, and PF is the power factor.

[0019] Deploy environmental data collection points on the wind turbine tower, install temperature sensors, humidity sensors and wind speed sensors to collect the temperature, humidity and wind speed data at the location of the wind turbine generator set.

[0020] Preprocess the collected historical power generation data, including removing outliers, missing values and normalization processing, calculate the Z - score of each parameter in the historical power generation data. When the absolute value of the Z - score of the parameter |Z|>3, the parameter is regarded as an outlier and is removed. After removing the outliers, the missing values of the parameters are filled with the mean value of the previous moment and the next moment parameters; after the outlier and missing value processing, the Min - Max method is used for normalization processing to map all parameters to the [0,1] interval. The historical power generation data after preprocessing is sorted into a historical data set according to the time stamp.

[0021] Further, the specific process of training the multiple linear regression model includes:

[0022] Take the actual power, generator speed, wind speed, temperature and humidity data at the previous moment in the historical data set as the input, take the actual power at the next moment as the label, select the multiple linear regression model, and express the multiple linear regression model as:

[0023] P(t)=β0·+β1·N(t - 1)+β2·P(t - 1)+β3·V(t - 1)+β4T(t - 1)+β5·H(t)+∈

[0024] Wherein, P(t) is the actual power at time t, N(t - 1) is the rotational speed of the wind turbine at time t - 1, P(t - 1) is the actual power at time t - 1, V(t - 1) is the wind speed at time t - 1, T(t - 1) is the temperature at time t - 1, H(t - 1) is the humidity at time t - 1, β0 is the intercept, β1, β2, β3, β4, β5 are the influence coefficients of each parameter, t is the time variable, and ∈ is the error term;

[0025] The historical data set is divided into a training set and a test set, with 80% used for training and 20% used for testing. The data in the training set is used to fit the regression model. The mean squared error is selected as the loss function, and the least squares method is used to minimize the loss function, thereby determining each regression coefficient. Monitor the root mean squared error during the training process in the training model. When the decrease amplitude is lower than 0.01 within 10 consecutive rounds, stop the training, indicating that the model training is completed. After the model training is completed, use the test set for model evaluation and calculate the coefficient of determination R 2 value to evaluate the model performance.

[0026] Further, training the power prediction model specifically includes:

[0027] Set the input sequence length of the long short - term memory network to m time steps, where m takes 10. Input the historical power generation data of the first 10 moments in the training set, and the actual power data of the next moment in the training set as the label. Select the long short - term memory network model to train the power prediction model. After the training is completed, input the test set into the trained power prediction model for testing. Select the mean squared error as the loss function. When the decrease amplitude of the validation set loss is lower than 0.001 within 10 consecutive rounds, it indicates that the training of the power quality prediction model is completed.

[0028] Further, the calculation of the fluctuation adjustment coefficient specifically includes:

[0029] Arrange the generator speed, wind speed, temperature, and humidity data at the current moment and the previous m - 1 moments in the order of acquisition time to form generator speed time - series data, wind speed time - series data, temperature time - series data, and humidity time - series data. Calculate the fluctuation index of the generator speed as:

[0030]

[0031] Where, W N is the fluctuation index of the generator speed, σ N is the standard deviation of the generator speed time - series data, and μ N is the mean of the generator speed time - series data;

[0032] Similarly, the fluctuation indices of wind speed, temperature, and humidity can be obtained, which are W respectivelyV , W T , W H ;

[0033] Analyze the fluctuation indicators of wind speed, temperature, and humidity to obtain natural indicators. The calculation formula is:

[0034] E = ω V ·W V + ω T ·W T + ω H ·W H

[0035] Among them, E is the natural indicator, ω V , ω T , ω H are the weight coefficients of wind speed, temperature, and humidity in the natural indicator respectively. 0 < ω H < ω T < W V < 1, and ω V + ω T + ω H = 1;

[0036] The formula for calculating the fluctuation adjustment coefficient is expressed as:

[0037]

[0038] Among them, α is the fluctuation adjustment coefficient. Further, the actual power after correcting the calculated fluctuation adjustment coefficient specifically includes:

[0039] Collect the power generation data of the wind turbine at the current moment, including generator speed, wind speed, temperature, humidity, and the actual power data at the current moment. Input the power generation data of the previous multiple moments and the current moment into the trained power prediction model to obtain the actual power predicted by the model. Then, correct the actual power predicted by the multiple linear regression model and the power prediction model through the fluctuation adjustment coefficient. The formula is expressed as:

[0040]

[0041] Among them, is the actual power after correcting by the fluctuation adjustment coefficient, is the actual power predicted by the output of the multiple linear regression model, is the actual power predicted by the output of the power prediction model;

[0042] Apply the actual power after correcting the fluctuation adjustment coefficient to the power curve of the wind turbine at the next moment, and compare the actual power after correcting the fluctuation adjustment coefficient with a preset power threshold. If the actual power after correcting the fluctuation adjustment coefficient is lower than the preset power threshold, the wind turbine continues to monitor. If the actual power after correcting the fluctuation adjustment coefficient is higher than the preset power threshold, an abnormal warning is issued.

[0043] The present invention further provides a device for predicting the power curve of a wind turbine. The device for predicting the power curve of a wind turbine is used to implement the above-mentioned method for predicting the power curve of a wind turbine, and includes:

[0044] A historical data acquisition module, configured to acquire historical power generation data of the wind power generation unit. The historical power generation data includes the actual power, generator speed, and environmental data at each moment in the power curve of the generator set. The environmental data includes wind speed, temperature, and humidity;

[0045] A regression model construction module, configured to use the actual power, generator speed, and environmental data at the previous moment in the historical power generation data as inputs, and the actual power at the next moment as a label, to construct and train a multiple linear regression model;

[0046] A power prediction model construction module, configured to use the historical power generation data at the previous m moments in the historical power generation data as inputs, and the actual power at the next moment as a label, to construct and train a power prediction model based on a long short-term memory network model;

[0047] A fluctuation adjustment coefficient calculation module, configured to calculate the fluctuation adjustment coefficient at the current moment according to the generator speed, wind speed, temperature, and humidity data at the current moment;

[0048] A real-time data prediction and correction module, configured to acquire the power generation data at the current moment and the previous m - 1 moments, input them into the multiple linear regression model and the power prediction model, obtain the actual power at the next moment predicted by the two models, and then calculate the actual power after correcting the fluctuation adjustment coefficient in combination with the fluctuation adjustment coefficient, and apply the corrected actual power to the power curve of the wind turbine. In the above technical solution, the technical effects and advantages provided by the present invention are:

[0049] The present invention obtains the influence coefficients of various parameters in historical power generation data on the power of a wind turbine through multiple linear regression analysis of the historical power generation data, and predicts the actual power at future moments by constructing a Long Short-Term Memory (LSTM) model, accurately capturing the influence of the generator speed and environmental data on the actual power, thereby improving the accuracy and reliability of the prediction. By introducing a fluctuation adjustment coefficient to correct the actual power predicted by the two models, the final prediction result is closer to the actual power generation situation, enabling real-time monitoring and issuing abnormal warnings, providing strong support for the scheduling and management of wind power generation, and thus promoting the efficient utilization and stability of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a schematic flow chart of the overall method of the present invention;

[0051] Figure 2 is a schematic structural diagram of the device of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0053] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs. The "first", "second", and similar terms used in the present invention do not denote any order, quantity, or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0054] Embodiment:

[0055] Please refer to Figure 1 , the present invention provides a technical solution:

[0056] A method for predicting the power curve of a wind turbine generator set, the specific steps including:

[0057] Step 1: Collect the historical power generation data of the wind power generation set, where the historical power generation data includes the actual power, generator speed, and environmental data at each moment in the power curve of the generator set, and the environmental data includes wind speed, temperature, and humidity;

[0058] In this embodiment, the specific process of collecting historical power generation data of the wind turbine generator set includes:

[0059] Collect the current, voltage and generator speed data of the wind turbine generator set through the power sensors and speed sensors equipped on the wind turbine generator set, and obtain the power factor of the wind turbine generator set through a computing watt-hour meter. The formula for calculating the actual power of the wind turbine generator set is expressed as:

[0060]

[0061] where P is the actual power, V is the voltage, I is the current, and PF is the power factor. Here, the calculation of the actual power represents a three-phase system;

[0062] Deploy environmental data collection points on the wind turbine tower, and install temperature sensors, humidity sensors and wind speed sensors to collect the temperature, humidity and wind speed data at the location of the wind turbine generator set;

[0063] Preprocess the collected historical power generation data, including removing outliers, missing values and normalization processing. Calculate the Z-score of each parameter in the historical power generation data. When the absolute value of the Z-score of the parameter |Z|>3, the parameter is regarded as an outlier and removed. After removing the outliers, the missing values are filled with the mean value of the previous and next moment parameters; after the outlier and missing value processing, the Min-Max method is used for normalization processing to map all parameters to the [0,1] interval. The historical power generation data after preprocessing is sorted into a historical data set according to the time stamp.

[0064] By collecting the historical power generation data of the wind turbine generator set, covering the actual power, speed and environmental data (wind speed, temperature and humidity) of the generator set, a comprehensive database is provided. These data are crucial for understanding the operating characteristics of the wind turbine generator set and its response to environmental changes. Deploying environmental data collection points on the wind turbine tower can timely obtain wind speed, temperature and humidity data. This real-time monitoring method combined with the operating state of the generator set increases the response ability to external environmental changes and more accurately reflects the impact of environmental factors on power generation performance. By detecting and removing outliers through the Z-score method, the reliability of the data is ensured, and the missing values are filled with the mean value to avoid the instability of model training caused by data loss. This process improves the integrity and quality of the data set and provides a more robust data basis for subsequent model training. Through Min-Max normalization processing, all parameters are mapped to the [0,1] interval, which not only eliminates the influence between different dimensions, but also improves the convergence speed of model training and the stability of prediction, enabling different features to be compared and analyzed on the same scale.

[0065] Step 2: Use the actual power, generator speed, and environmental data at the previous moment in the historical power generation data as inputs, and the actual power at the next moment as the label to construct and train a multiple linear regression model.

[0066] In this embodiment, training the multiple linear regression model specifically includes:

[0067] Use the actual power, generator speed, wind speed, temperature, and humidity data at the previous moment in the historical dataset as inputs, and the actual power at the next moment as the label. Select a multiple linear regression model and represent the multiple linear regression model as:

[0068] P(t) = β0 + β1·N(t - 1) + β2·P(t - 1) + β3·V(t - 1) + β4·T(t - 1) + β5·H(t) + ∈

[0069] Where P(t) is the actual power at time t, N(t - 1) is the speed of the wind turbine at time t - 1, P(t - 1) is the actual power at time t - 1, V(t - 1) is the wind speed at time t - 1, T(t - 1) is the temperature at time t - 1, H(t - 1) is the humidity at time t - 1, β0 is the intercept, β1, β2, β3, β4, β5 are the influence coefficients of each parameter, t is the time variable, and ∈ is the error term.

[0070] Divide the historical dataset into a training set and a test set, with 80% for training and 20% for testing. Use the data in the training set to fit the regression model. Select the mean squared error as the loss function and use the least squares method to minimize the loss function to determine each regression coefficient. Monitor the root mean squared error during the training process in the training model. When the decrease amplitude is less than 0.01 within 10 consecutive rounds, stop training, indicating that the model training is completed. After the model training is completed, use the test set to evaluate the model and calculate the coefficient of determination R 2 value to evaluate the model performance. The smaller the R 2 value, the better the model performance.

[0071] By selecting the generator speed and environmental data (wind speed, temperature, humidity) as features, using the actual power at the previous moment as the input, and the target is the actual power at the current moment. This selection conforms to the physical characteristics of wind turbines, can accurately capture the key factors affecting power generation, and helps the effectiveness of the model. The multiple linear regression model is a classic statistical method suitable for dealing with the linear relationship between features and targets. By constructing this model, the impact degree of each feature on the actual power can be clearly understood, which helps to optimize the operation and management of wind turbines. The historical dataset is divided into a training set and a test set (80% for training and 20% for testing). This division method ensures the generalization ability of the model on different datasets. The mean square error is used as the loss function during training, and the least squares method is used to optimize the regression coefficients to ensure that the model can effectively fit the historical data. In addition, by monitoring the root mean square error (RMSE) and setting the conditions for stopping training, overfitting is avoided and the robustness of the model is improved. Use the R 2 value to evaluate the model performance. The closer the R 2 value is to 1, the better the model fits the data. Appropriate evaluation criteria help to quantify the model performance and ensure that the final model can accurately and reliably predict the power generation.

[0072] Step 3: Use the historical power generation data of the previous m moments in the historical power generation data as the input, and the actual power of the next moment as the label. Based on the long short-term memory network model, construct and train a power prediction model;

[0073] In this embodiment, training the power prediction model specifically includes:

[0074] Set the input sequence length of the long short-term memory network to m time steps, where m is 10. Input the historical power generation data of the first 10 moments in the training set, and the actual power data of the next moment in the training set as the label. Select the long short-term memory network model to train the power prediction model. After training, input the test set into the trained power prediction model for testing. Select the mean square error as the loss function. When the loss of the validation set drops by less than 0.001 in 10 consecutive rounds, it marks the completion of the training of the power quality prediction model. Here, the training set and the test set are the training set and the test set divided in Step 2.

[0075] The Long Short-Term Memory Network (LSTM) is a deep learning model specifically designed to process sequential data and can effectively capture long-term dependencies in time series. By taking the historical power generation data at the previous t-1 time steps as input, the LSTM can learn the patterns implicit in the historical data, thereby enhancing the model's prediction ability. When processing sequential data, the LSTM network can effectively overcome the problems of gradient vanishing and explosion in traditional Recurrent Neural Networks (RNNs) during long sequence learning. This characteristic makes the LSTM very suitable for wind power prediction because it can capture both long-term and short-term information related to time. Selecting 10 time steps as the input length is a reasonable choice because it can capture the changing trends of factors such as wind speed, temperature, and rotational speed in a short period. According to the actual characteristics of the data and the training effect of the model, this parameter can be further optimized, but the choice of 10 time steps usually performs well in practice. Selecting the Mean Squared Error (MSE) as the loss function can effectively measure the difference between the model's predicted value and the actual value. During training, setting the condition that the loss of the validation set drops by less than 0.001 for 10 consecutive epochs as the stopping condition ensures that the model can achieve good convergence and generalization ability during training, thereby improving the accuracy and robustness of the prediction.

[0076] Step 4: Calculate the fluctuation adjustment coefficient at the current moment based on the generator rotational speed, wind speed, temperature, and humidity data at the current moment.

[0077] In this embodiment, the calculation of the fluctuation adjustment coefficient specifically includes:

[0078] Arrange the generator rotational speed, wind speed, temperature, and humidity data at the current moment and the previous m-1 time steps in the order of collection time to form generator rotational speed time series data, wind speed time series data, temperature time series data, and humidity time series data. Calculate the fluctuation index of the generator rotational speed as:

[0079]

[0080] where, W N is the fluctuation index of the generator rotational speed, σ N is the standard deviation of the generator rotational speed time series data, and μ N is the mean of the generator rotational speed time series data;

[0081] Similarly, the fluctuation indices of wind speed, temperature, and humidity can be obtained, which are W C , W T , W H ;

[0082] Analyze the fluctuation indices of wind speed, temperature, and humidity to obtain the natural index. The calculation formula is:

[0083] E = ωV ·W V + ω T ·W T + ω H ·W H

[0084] where E is the natural index, and ω V , ω T , ω H are the weight coefficients of wind speed, temperature, and humidity in the natural index, respectively, where 0 < ω H < ω T < ω V < 1, and ω V + ω T + W H = 1;

[0085] The natural index is a quantity that comprehensively considers the fluctuations of wind speed, temperature, and humidity. Its purpose is to quantify the degree of influence of environmental factors on the power output of wind turbines. Wind speed, temperature, and humidity are the key environmental factors that affect the operating efficiency and power output of wind turbines. Different environmental conditions will have a significant impact on the extraction efficiency of wind energy. Therefore, the fluctuations of these factors need to be taken into account. Wind speed is one of the most critical factors in wind power generation. The change in wind speed directly affects the availability of wind energy. Therefore, its fluctuation index has a relatively large impact on the generated power. Due to the huge impact of wind speed on the generated power, the greater the wind speed, the larger its natural index, and the larger the fluctuation adjustment coefficient. This indicates that when the environmental fluctuation is large, it is more dependent on the power prediction model. Therefore, its weight is the largest. Temperature affects the efficiency and performance of wind turbines, including air density, and thus affects the wind energy conversion efficiency. Although the change in temperature has a smaller impact on the power output than wind speed, it is still important. Its degree of influence is lower than that of wind speed. Therefore, its weight is set lower than that of wind speed. Humidity has a relatively small impact on wind power generation. Only under specific circumstances (such as a high-humidity environment) may it affect the operating conditions of the equipment. Among these three environmental factors, humidity has the relatively smallest impact on the natural index. Therefore, its weight is set to be the smallest.

[0086] The formula for calculating the fluctuation adjustment coefficient is expressed as:

[0087]

[0088] where α is the fluctuation adjustment coefficient.

[0089] When the environmental fluctuation is large, this means that the change in the environment has a significant impact on the power. At this time, the value of α will be high, and it is more dependent on the power prediction model; when the generator speed is large, this means that the system is in a relatively unstable state, and the impact at the current moment is large. At this time, the value of α will be small, and it is more dependent on the multiple linear regression model.

[0090] The power generation capacity of a wind turbine is affected by various environmental factors, including wind speed, temperature, humidity, etc. The fluctuations of these factors can be significant under different times and conditions. By introducing a fluctuation adjustment coefficient α, the prediction model can adjust the corresponding weights in real time according to the current environmental conditions, thereby enhancing the adaptability of the model to different working environments. In practical applications, the operating state and environmental conditions of a wind turbine are not always stable. When the environment fluctuates greatly (such as a drastic change in wind speed), the traditional multiple linear regression model may not be able to capture the impact brought by these changes well. On the contrary, the long short-term memory network (LSTM) has advantages in processing time series data and capturing dynamic patterns. By appropriately reducing the weight of the multiple linear regression model (i.e., increasing the weight of the LSTM model), more accurate predictions can be obtained under high-fluctuation conditions. Each model has its limitations. The multiple linear regression model assumes a linear relationship between the input and output and is sensitive to outliers; while the LSTM model, although able to handle complex non-linear relationships, may also perform poorly when the data volume is insufficient or the feature selection is inappropriate. Implementing the steps of the fluctuation adjustment coefficient can effectively combine the advantages of the two models, select a more suitable model under different environmental conditions, and overcome the limitations of a single model. The introduction of the fluctuation adjustment coefficient can help reduce the prediction error caused by changes in environmental factors. When the fluctuation is large, the system can preferentially rely on the LSTM model that is sensitive to dynamic changes to reduce the prediction inaccuracy caused by drastic environmental changes; while in a relatively stable environment, relying on the multiple linear regression model can provide more stable prediction results. This flexibility reduces the overall prediction error.

[0091] Step 5: Collect the power generation data at the current moment and the previous m - 1 moments, input them into the multiple linear regression model and the power prediction model, obtain the actual power at the next moment predicted by the two models, and then calculate the actual power corrected by the fluctuation adjustment coefficient in combination with the fluctuation adjustment coefficient, and apply the corrected actual power to the power curve of the wind turbine;

[0092] In this embodiment, the calculation of the actual power corrected by the fluctuation adjustment coefficient specifically includes:

[0093] Collect the power generation data of the wind turbine at the current moment, including the generator speed, wind speed, temperature, humidity, and the actual power data at the current moment. Input the power generation data at the previous multiple moments and the current moment into the trained power prediction model to obtain the actual power predicted by the model, and then correct the actual power predicted by the multiple linear regression model and the power prediction model through the fluctuation adjustment coefficient. The formula is expressed as:

[0094]

[0095] Where, is the actual power after being corrected by the fluctuation adjustment coefficient, is the actual power predicted by the output of the multiple linear regression model, is the actual power predicted by the output of the power prediction model;

[0096] Apply the actual power after being corrected by the fluctuation adjustment coefficient to the power curve of the wind turbine at the next moment, and compare the actual power after being corrected by the fluctuation adjustment coefficient with a preset power threshold. If the actual power after being corrected by the fluctuation adjustment coefficient is lower than the preset power threshold, the wind turbine continues to monitor. If the actual power after being corrected by the fluctuation adjustment coefficient is higher than the preset power threshold, an abnormal warning is issued.

[0097] The multiple linear regression model (usually performs well when the environmental conditions do not change much and relies on the linear relationship of historical data. The power prediction model (such as deep learning models like LSTM) is more adaptable to dynamic and complex environmental changes and can capture more complex non-linear relationships. Therefore, by weighted fusion of the outputs of these two models, the respective advantages can be fully utilized to improve the prediction accuracy. The introduction of the fluctuation adjustment coefficient α is to dynamically adjust the weights of the two models to adapt to different environmental fluctuation states. When the environmental fluctuation is large, α is large, making the weight of the power prediction model larger, enhancing the flexibility and accuracy of the prediction. When the environmental fluctuation is small, α is small, increasing the stability of the prediction. This mechanism can meet the prediction requirements under different environmental conditions such as wind speed, temperature, and humidity, achieving more accurate power prediction.

[0098] Please refer to Figure 2 , the present invention further provides a device for predicting the power curve of a wind turbine. The device for predicting the power curve of the wind turbine is used to implement the above-mentioned method for predicting the power curve of the wind turbine, and includes:

[0099] A historical data acquisition module, configured to acquire historical power generation data of the wind power generation group. The historical power generation data includes the actual power, generator speed, and environmental data at each moment in the power curve of the generator set. The environmental data includes wind speed, temperature, and humidity;

[0100] A regression model construction module, configured to use the actual power, generator speed, and environmental data at the previous moment in the historical power generation data as inputs, and the actual power at the next moment as a label to construct and train a multiple linear regression model;

[0101] A power prediction model construction module, configured to use the historical power generation data of the previous m moments in the historical power generation data as inputs, and the actual power at the next moment as a label to construct and train a power prediction model based on a long short-term memory network model;

[0102] A fluctuation adjustment coefficient calculation module, which is used to calculate the fluctuation adjustment coefficient at the current moment according to the generator speed, wind speed, temperature, and humidity data at the current moment;

[0103] A real-time data prediction and correction module, which is used to collect the power generation data at the current moment and the previous m - 1 moments, input them into the multiple linear regression model and the power prediction model, obtain the actual power at the next moment predicted by the two models, and then calculate the actual power after correction of the fluctuation adjustment coefficient in combination with the fluctuation adjustment coefficient, and apply the corrected actual power to the power curve of the wind turbine generator.

[0104] The above formulas are all calculated by taking the numerical values after dimensionless. The formula is a formula obtained by software simulation of collecting a large amount of data to approximate the real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or by the combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0106] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0107] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application.

Claims

1. A method for predicting the power curve of a wind turbine generator set, characterized in that, The specific steps include: Step 1: Collect historical power generation data of the wind power generation group, the historical power generation data includes the actual power, generator speed and environmental data at each moment in the power curve of the generator group, and the environmental data includes wind speed, temperature and humidity; Step 2: Take the actual power, generator speed and environmental data of the previous moment in the historical power generation data as input, and the actual power of the next moment as the label, and build and train a multivariate linear regression model; Step 3: Take the historical power generation data of the previous m moments as input and the actual power at the next moment as the label, and build and train the power prediction model based on the long short-term memory network model; Step 4: Calculate the fluctuation adjustment coefficient at the current moment based on the generator speed, wind speed, temperature and humidity data at the current moment; Step 5: Collect the power generation data at the current moment and the previous m-1 moments, input them into the multivariate linear regression model and the power prediction model, obtain the actual power at the next moment predicted by the two models, and then calculate the actual power corrected by the fluctuation adjustment coefficient in combination with the fluctuation adjustment coefficient, and apply the corrected actual power to the power curve of the wind turbine generator set.

2. A method for predicting the power curve of a wind turbine generator set according to claim 1, characterized in that, The collection of historical power generation data of wind turbine generator sets specifically includes: The power sensor and speed sensor equipped with the wind turbine are used to collect the current, voltage and generator speed data of the wind turbine. The power factor of the wind turbine is obtained through the calculation type electric energy meter. The formula for calculating the actual power of the wind turbine is expressed as follows: Among them, P is the actual power, V is the voltage, I is the current, and PF is the power factor; Deploy environmental data collection points on wind turbine towers and install temperature sensors, humidity sensors, and wind speed sensors to collect temperature, humidity, and wind speed data at the wind turbine locations; The collected historical power generation data are preprocessed, including removing outliers, missing values and normalizing. The Z-score of each parameter in the historical power generation data is calculated. When the absolute value of the Z-score of the parameter |Z|>3, the parameter is regarded as an outlier and removed. After the outliers are removed, the parameters use the mean of the parameters at the previous moment and the next moment to fill the missing values. After the outliers and missing values are processed, the Min-Max method is used for normalization, and all parameters are mapped to the [0,1] interval. After the preprocessing, the historical power generation data is organized into a historical data set according to the timestamp.

3. A method for predicting the power curve of a wind turbine generator set according to claim 1, characterized in that, Training the multivariate linear regression model specifically includes: The actual power, generator speed, wind speed, temperature and humidity data of the previous moment in the historical data set are taken as input, the actual power of the next moment is taken as the label, and the multiple linear regression model is selected. The multiple linear regression model is expressed as: P(t)=β0·+β1·N(t-1)+β2·P(t-1)+β3·V(t-1)+β4·T(t-1)+β5·H(t)+∈ Wherein, P(t) is the actual power at time t, N(t - 1) is the rotational speed of the wind turbine at time t - 1, P(t - 1) is the actual power at time t - 1, V(t - 1) is the wind speed at time t - 1, T(t - 1) is the temperature at time t - 1, H(t - 1) is the humidity at time t - 1, β0 is the intercept, β1, β2, β3, β4, β5 are the influence coefficients of each parameter, t is the time variable, and ∈ is the error term; Divide the historical dataset into a training set and a test set, with 80% for training and 20% for testing. Use the data in the training set to fit a regression model. Select the mean squared error as the loss function and use the least squares method to minimize the loss function, thereby determining the regression coefficients. Monitor the root mean squared error during the training process in the training model. When the decrease amplitude is less than 0.01 within 10 consecutive rounds, stop the training, indicating that the model training is completed. After the model training is completed, use the test set for model evaluation and calculate the coefficient of determination R 2 value to evaluate the model performance.

4. A method for predicting the power curve of a wind turbine generator set according to claim 3, characterized in that, Specifically, training the power prediction model includes: Setting the input sequence length of the long short-term memory network to m time steps, where m is taken as 10. Inputting the historical power generation data of the first 10 moments in the training set, and using the actual power data of the next moment in the training set as the label. Selecting the long short-term memory network model to train the power prediction model. After training is completed, input the test set into the trained power prediction model for testing. Selecting the mean square error as the loss function. When the decrease in the loss of the validation set is less than 0.001 within 10 consecutive rounds, it indicates that the training of the power quality prediction model is completed.

5. A method for predicting the power curve of a wind turbine generator set according to claim 1, characterized in that Specifically, calculating the fluctuation adjustment coefficient includes: Arranging the generator speed, wind speed, temperature, and humidity data of the current moment and the previous m - 1 moments in the order of collection time to form generator speed time series data, wind speed time series data, temperature time series data, and humidity time series data. The calculation formula for the fluctuation index of the generator speed is: Among them, W N is the fluctuation index of the generator speed, and σ N is the standard deviation of the generator speed time series data, and μ N is the mean value of the generator speed time series data; Similarly, the fluctuation indices of wind speed, temperature, and humidity can be obtained, which are W V , W T , and W H ; Analyzing the fluctuation indexes of wind speed, temperature, and humidity to obtain the natural index. The calculation formula is: E = ω V ·W V + ω T ·W T + ω H ·W H where E is the natural index, ω V , ω T , ω H are the weight coefficients of wind speed, temperature, and humidity in the natural index, respectively, 0 < ω H < ω T < ω V < 1, and ω V + ω T + ω H = 1; The calculation formula for the fluctuation adjustment coefficient is expressed as: Where α is the fluctuation adjustment coefficient.

6. A method for predicting the power curve of a wind turbine generator set according to claim 1, characterized in that Specifically, calculating the actual power corrected by the fluctuation adjustment coefficient includes: Collecting the power generation data of the current moment of the wind turbine generator set, including the generator speed, wind speed, temperature, humidity, and the actual power data of the current moment. Inputting the power generation data of the previous multiple moments and the current moment into the trained power prediction model to obtain the actual power predicted by the model. Then, correcting the actual power predicted by the multiple linear regression model and the power prediction model through the fluctuation adjustment coefficient. The formula is expressed as: Among them, is the actual power after correction by the fluctuation adjustment coefficient, is the actual power predicted by the output of the multiple linear regression model, is the actual power predicted by the output of the power prediction model; Applying the actual power corrected by the fluctuation adjustment coefficient to the power curve of the wind turbine generator set at the next moment, and comparing the actual power corrected by the fluctuation adjustment coefficient with the preset power threshold. If the actual power corrected by the fluctuation adjustment coefficient is lower than the preset power threshold, the wind turbine generator set continues to be monitored. If the actual power corrected by the fluctuation adjustment coefficient is higher than the preset power threshold, an abnormal warning is issued.

7. A device for predicting the power curve of a wind turbine generator, characterized in that The power curve prediction device of the wind turbine generator set is used to implement the power curve prediction method of the wind turbine generator set according to any one of claims 1 - 6, and includes: A historical data acquisition module, which is used to acquire the historical power generation data of the wind power generation set. The historical power generation data includes the actual power, generator speed, and environmental data at each moment in the power curve of the generator set. The environmental data includes wind speed, temperature, and humidity; A regression model construction module, which is used to take the actual power, generator speed, and environmental data of the previous moment in the historical power generation data as inputs, and the actual power of the next moment as the label, to construct and train a multiple linear regression model; The power prediction model construction module is used to take the historical power generation data of the previous m moments in the historical power generation data as input, and the actual power of the next moment as the label, and construct and train a power prediction model based on the long short-term memory network model; The fluctuation adjustment coefficient calculation module is used to calculate the fluctuation adjustment coefficient at the current moment according to the generator speed, wind speed, temperature and humidity data at the current moment; The real-time data prediction and correction module is used to collect the power generation data at the current moment and the previous m-1 moments, input them into the multiple linear regression model and the power prediction model, obtain the actual power of the next moment predicted by the two models, and then combine the fluctuation adjustment coefficient to calculate the actual power after the fluctuation adjustment coefficient is corrected, and apply the corrected actual power to the power curve of the wind turbine generator.

Images