Fan power grid management system based on wind speed prediction and adaptive control strategy

Abstract

The invention discloses a fan power grid management system based on wind speed prediction and a self-adaptive control strategy, and relates to the technical field of wind power generation. During wind speed prediction, a wind turbine generator is divided into a plurality of clusters through KDBSCAN, each fan in the clusters is subjected to wind speed prediction model training by utilizing HOT-RNN, and the models are combined, so that the wind speed prediction is realized; a threshold value corresponding to each prediction model is obtained through an optimization algorithm, a kernel ridge regression model is introduced to find out a nonlinear function relation between a prediction value and a prediction error, a total wind speed prediction model is obtained, the prediction model considers nonlinear conditions of a plurality of fans and actual wind speed data, and wind speed prediction is more accurate. Meanwhile, for the moment when the wind speed fluctuates greatly, the self-adaptive attack angle increment and the self-adaptive rotating speed adjusting rate are introduced, so that each fan can convert electric energy to the maximum extent according to the actual change of the wind speed, and the safety of the equipment is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind turbine power grid management system based on wind speed prediction and adaptive control strategy. Background Art

[0002] In the field of wind power generation, the uncertainty of wind speed and the coordination problems between wind turbines and power grids have always plagued the development of the industry. Especially in terms of wind speed prediction, the traditional prediction method only predicts the wind speed of a certain wind turbine of the wind turbine group. However, due to the influence of factors such as the location and environment of each wind turbine, the prediction effect of only one is not good, and the wind speed data itself is extremely nonlinear. It is also extremely important to choose a suitable wind speed prediction model. At the same time, traditional wind turbine control methods are difficult to accurately adapt to wind speed changes, resulting in low wind turbine power generation efficiency and impact on the stability of the power grid. For example, when the wind speed changes suddenly, the output power of the wind turbine fluctuates violently, which is easy to exceed the acceptance capacity of the power grid and cause power grid failure; and inaccurate wind speed prediction makes it impossible to reasonably deploy energy storage equipment, resulting in energy waste or insufficient power supply. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides a wind turbine power grid management system based on wind speed prediction and adaptive control strategy, which solves the problems of inaccurate wind speed prediction for a single wind turbine and difficulty in coordinating the wind turbine and the power grid.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: A wind turbine power grid management system based on wind speed prediction and adaptive control strategy, comprising:

[0005] In the wind speed prediction module, the KDBSCAN density clustering algorithm is used to divide the wind turbine group into n clusters. Each cluster has Vc (c∈[1,n]) similar wind turbines. HOT-RNN is used to train each wind turbine in the n clusters, and the trained Vc similar wind turbines in each cluster are combined. The corresponding weights are obtained through the optimization algorithm. At the same time, the kernel ridge regression model is introduced to determine the relationship between the prediction error and the prediction value, and the final wind speed prediction model is constructed.

[0006] The energy storage control module continuously monitors the wind speed state through the wind speed prediction module, calculates the wind speed change k at adjacent moments, and activates the energy storage device if |k|>d, and does not activate the energy storage device if |k|<=d, where d is the wind speed change threshold;

[0007] Wind turbine attack angle adaptive control module, if |k|>d and k>0, adjust the wind turbine blade attack angle according to the adaptive attack angle increment, if P(t)>Pg(t), activate the energy storage device for charging, where P(t) is the wind turbine power generation at time t, and Pg(t) is the stable power that the power grid can accept at time t;

[0008] The wind turbine speed adaptive control module, if |k|>d and k<0, adjusts the wind turbine speed according to the adaptive speed adjustment rate. If Pr(t)>P(t), activates the energy storage device for discharge, where Pr(t) is the real-time power demand of the power grid.

[0009] As a further solution of the present invention, the specific steps of using the KDBSCAN density clustering algorithm to divide the wind turbines include:

[0010] Collect multi-dimensional data of wind turbines, including wind speed v, power p, longitude lon, latitude lat, generator speed g, and construct a data set X = {x1, x2, ..., xm}, xi = (vi, pi, loni, lati, gi), where m is the number of wind turbines in the wind turbine;

[0011] Perform Z-score standardization on each xi in the data set. If there are missing values, fill them with the mean value.

[0012] Determine the core point neighborhood radius eps according to the scale and distribution density of wind turbines, and set the minimum number of points min_samples;

[0013] According to the formula Calculate the number of points in the eps neighborhood of point xi. If Neps(xi)>=min_samples, then xi is a core point, where d(xi,xj) is the Euclidean distance between two points, and I(·) is the indicator function, which is 1 if satisfied, otherwise 0;

[0014] Starting from the core point xi, if xj is in the eps neighborhood of xi, xj and xi are grouped into one cluster;

[0015] According to the formula Calculate the silhouette coefficient S to evaluate the clustering effect, and adjust the parameters accordingly to optimize the clustering, where ai is the average distance from xi to other points in the same cluster, and bi is the average distance from xi to all points in the nearest neighbor cluster.

[0016] As a further solution of the present invention, the specific method of using HOT-RNN to train the wind turbine is:

[0017] Collect a large amount of historical operating data from the wind turbine monitoring system, including wind speed, wind direction, power output, humidity, ambient temperature, and air pressure, to construct a multivariate raw data set;

[0018] Perform Min-Max normalization and data smoothing operations on the original data;

[0019] The preprocessed data is divided into a training set and a test set at 80% and 20%, and the divided multivariate time series data is converted into a high-order tensor form;

[0020] Determine the network architecture parameters, including the number of hidden layers, the number of neurons in each layer, the tensor order, and the time step;

[0021] Based on the training set data, the mean square error (MSE) and mean absolute error (MAE) are used to measure the deviation between the predicted value and the true value, and Adam is used to iteratively update the model parameters;

[0022] Substitute the test set data into the trained prediction model and calculate the MSE, MAE, and prediction accuracy to comprehensively measure the prediction accuracy, stability, and reliability of the model. If the requirements are not met, return to the above steps to continue training the model.

[0023] As a further solution of the present invention, the optimization algorithm includes a goose flock optimization algorithm, a black kite optimization algorithm, a genetic optimization algorithm, a simulated annealing optimization algorithm, and a particle swarm optimization algorithm.

[0024] As a further solution of the present invention, the specific steps of introducing the kernel ridge regression model to determine the relationship between the prediction error and the predicted value are:

[0025] Collect the predicted values ​​of the Vc group prediction model as A1p, A2p, ..., AVcp and the corresponding error dp to build a data set Among them, g is the total amount of predicted data;

[0026] According to the formula Map the input data to a high-dimensional feature space and calculate the kernel matrix K, where xp = (A1 p, A2p, ..., AVcp), xq = (A1 q, A2q, ..., AVcq), and σ is the kernel bandwidth parameter;

[0027] Let the regression coefficient vector be a=(a0,a1,...,aVc) T ,Depend on As the minimization objective function, where λ is the regularization parameter;

[0028] Take the partial derivative of J(a) with respect to a and set it to 0 to determine the regression coefficient a;

[0029] Based on the regression coefficient a, the functional relationship f between the prediction error and the predicted value is constructed:

[0030]

[0031] As a further solution of the present invention, the final wind speed prediction model constructed is w1A1+w2A2+...+wVcAVc+f(A1,A2,...,AVc), wherein w1,w2,...,wVc are weights calculated by the optimization algorithm.

[0032] As a further embodiment of the present invention, according to the formula Set the adaptive angle of attack increment, where ɑ is the current angle of attack, max is the maximum blade angle of attack, and m is the proportional coefficient.

[0033] As a further embodiment of the present invention, according to the formula To adjust the energy storage device for charging, where Ps(t) is the charging power of the energy storage device.

[0034] As a further embodiment of the present invention, according to the formula Get the fan adaptive speed adjustment rate, where R max is the maximum allowable speed change rate of the fan, v is the current wind speed, N(t) is the current fan speed, and n is a constant.

[0035] As a further embodiment of the present invention, according to the formula To adjust the energy storage device to discharge, where Pd(t) is the discharge power of the energy storage device.

[0036] The present invention provides a wind turbine power grid management system based on wind speed prediction and adaptive control strategy, which has the following beneficial effects compared with the prior art:

[0037] (1) The present invention clusters wind turbines in wind turbines by using the KDBSCAN density clustering algorithm, and can group wind turbines in similar operating states into one cluster. A wind speed prediction model is constructed for each cluster, and the training and optimization of common features within the cluster are focused on, so that the model is more in line with the actual working conditions. Compared with the overall unified modeling, the influence of random fluctuations in wind speed can be greatly reduced;

[0038] (2) The present invention predicts the wind speed of each wind turbine in each cluster through HOT-RNN, constructs the wind speed, wind direction, power output, humidity, ambient temperature, and air pressure into a high-order tensor input model, accurately captures the complex nonlinear correlation and dynamic change law of each factor, mines deep-level information, and comprehensively reflects the operating environment characteristics of the wind turbine, so that the wind speed prediction fully considers the comprehensive effect of multiple factors and improves the prediction accuracy;

[0039] (3) The present invention uses the kernel ridge regression model to determine the relationship between the prediction error and the predicted value. In the face of changes in wind speed fluctuation characteristics, the model can dynamically optimize the prediction value correction strategy to ensure that the wind speed prediction is close to the actual situation;

[0040] (4) The present invention introduces adaptive angle of attack increment and adaptive speed adjustment rate respectively for moments when wind speed fluctuates greatly, both of which can enable the wind turbine to convert electrical energy to the maximum extent according to the actual changes in wind speed and ensure the safety of the equipment itself. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a block diagram of the system principle of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] like Figure 1 The present application provides a wind turbine power grid management system based on wind speed prediction and adaptive control strategy, including:

[0044] The wind speed prediction module divides the wind turbines into n clusters using the KDBSCAN density clustering algorithm, and each cluster has Vc (c∈[1,n]) similar wind turbines:

[0045] Collect multi-dimensional data of wind turbines, including wind speed v, power p, longitude l on, latitude l at, and generator speed g, and construct a data set X = {x1, x2, ..., xm}, xi = (vi, pi, l on i, l at i, gi), where m is the number of wind turbines in the wind turbine;

[0046] Perform Z-score standardization on each xi in the data set, that is Among them, xij is the jth eigenvalue of the i-th sample, μ j and σ' j is the mean and standard deviation of the jth feature, making the mean of each feature 0 and the standard deviation 1 to avoid the influence of inconsistent dimensions on the clustering effect;

[0047] If there are missing values, the mean value can be used to fill them to ensure the integrity of the data;

[0048] It is initially determined based on the scale of wind turbines and the density of wind turbine distribution, such as eps = D / sqrt(sum(Vc)), where D is the diameter of the wind turbine geographical range, and sum(Vc) represents the total number of wind turbines in the wind turbine. It is then optimized based on the clustering effect. Considering data noise and the complexity of the expected cluster structure, the initial minimum number of points is set, such as min_samples = In(sum(Vc)), and then optimized based on the actual situation;

[0049] According to the formula Calculate the number of points in the eps neighborhood of point xi. If Neps(xi)>=min_samples, then xi is a core point, where d(xi,xj) is the Euclidean distance between two points, and I(·) is the indicator function, which is 1 if satisfied, otherwise 0;

[0050] Starting from the core point, clusters are constructed based on density reachability: if xj is in the eps neighborhood of the core point xi, then xj is density reachable from xi, and mutually density reachable points are grouped into one cluster;

[0051] According to the formula The silhouette coefficient S is calculated to evaluate the clustering effect. The mean of the silhouette coefficient reflects the overall clustering quality. A value close to 1 indicates good quality, and a value close to -1 indicates poor quality. The parameters are adjusted accordingly to optimize the clustering. Here, ai is the average distance from xi to other points in the same cluster, and bi is the average distance from xi to all points in the nearest neighbor cluster.

[0052] HOT-RNN is used to train each wind turbine in n clusters, including:

[0053] Various operating data of multiple wind turbines are collected from the monitoring system of the wind farm, including variables such as wind speed, wind direction, power, temperature, air pressure, and corresponding timestamp information. These data constitute the original multivariate time series data set, set as X = {(x1 t, x2t, ..., xqt), t∈[1, T]}, where q represents the number of variables, T represents the total length of the time series, and xit represents the observed value of the i-th variable at time step t;

[0054] For each variable i, we use Normalize it to the interval [0,1] and use the moving average method to smooth the normalized data to reduce the noise interference in the data. Usually, the smoothing window is set to r, and the calculation formula for the smoothed data is: For example, if the window size r = 5 is selected, the smoothing calculation of the wind speed data at time step t = 10 is to add the normalized wind speed values ​​from the 6th to the 10th time steps and take the average value as the smoothed wind speed value, which helps the model better capture the trend and periodic characteristics of the data instead of being misled by short-term noise fluctuations;

[0055] The preprocessed data is divided into a training set and a test set in a ratio of 8:2. The training set data is Xtrain = {(x1t, x2t, ..., xqt), t∈[1, Ttrai n]}, and the test set data is Xtest = {(x1t, x2t, ..., xqt), t∈[Trai n+1, T]}, where Ttrai n = 0.8T;

[0056] Convert the divided multivariate time series data into a high-order tensor form. The selection of the tensor order needs to be determined according to the multivariate and multi-time step characteristics of the data. Make full use of the structural information of the data so that the model can learn the interaction between variables and between different time steps. At the same time, it is also necessary to adjust the time step according to the periodicity of the data and the prediction requirements. For example, if it is found that the wind speed shows a relatively stable periodic feature within 24 time steps (assuming that each time step is 1 hour), the time step can be set to 24;

[0057] Taking the third-order tensor as an example, assuming that the time step is m1, the data can be organized into a tensor of shape (N, m1, q), where N is the number of samples (for the training set, N = Trai n / m1; for the test set, N = (T-Ttrai n) / m1). For example, for a data set containing three variables, wind speed, wind direction, and power, with a time step of 10, if there are a total of 1000 time steps of training data, then for the training set, N = (1000*0.8) / 10 = 80, and the shape of the tensor is (80, 10, 3). In this tensor, the first dimension represents different sample sequences, the second dimension represents the time step, and the third dimension represents the variable;

[0058] Determine the appropriate number of hidden layers based on the complexity of the data and the expressiveness requirements of the model. For wind speed prediction when the surrounding environment is stable, 2-3 hidden layers can be set. For prediction when the surrounding environment is complex and the wind speed fluctuates greatly, 3-5 hidden layers can be set.

[0059] The number of neurons in each layer also needs to be adjusted according to the characteristics of the data. For cases with higher dimensional input data (many variables or longer time steps), the number of neurons can be appropriately increased;

[0060] Based on the training set data, the mean square error (MSE) and mean absolute error (MAE) are used to measure the deviation between the predicted value and the true value. The Adam optimizer automatically adjusts the learning rate of each parameter according to the gradient changes of different parameters. For parameters with large gradient changes, the learning rate is appropriately reduced, and for parameters with small gradient changes, a relatively large learning rate is maintained, thereby improving the efficiency and stability of model training.

[0061] Load the training set data into the model in small batches for training. If the batch size is b, N / b batches of training are required in each training round. For each small batch, follow the above steps to train. After traversing all batches of data, a round of training is completed.

[0062] Substitute the test set into the trained prediction model, and calculate the evaluation indicators such as MSE, MAE, and prediction accuracy in the same way as the training process. If the performance of the model on the test set fails to meet the preset requirements, return to the above steps to continue training the model, which may involve the following operations, such as increasing or decreasing the number of hidden layers, adjusting the number of neurons, changing the tensor order or time step, or adjusting data preprocessing, such as modifying the normalization method, adjusting the window size for data smoothing, or adjusting the weight of the loss function or the optimizer parameters, until the prediction effect of the prediction model on the test set meets the preset standard.

[0063] Combine the trained Vc similar wind turbines in each cluster, and use the goose flock optimization algorithm, black kite optimization algorithm, genetic optimization algorithm, etc. to obtain their corresponding weights:

[0064] Determine the specific parameters required for the selected optimization algorithm. For example, for geese and black-winged kites, the population size N1 and the maximum number of iterations Tmax need to be set. For genetic algorithms, the crossover probability Pc and mutation probability Pm also need to be set. At the same time, determine the range of weights, set to [wmin, wmax], and randomly initialize the weight combination of each cluster Vc model within this range to form an initial population. Each individual can be represented as Li = (wi 1, wi2, ..., wiVc) (i∈[1, N1]);

[0065] For each individual Li in the current population, substitute it into the wind speed prediction model wi 1A1+wi2A2+...+wiVcAVc+f(A1,A2,...,AVc), and calculate the predicted wind speed value of the model in combination with the known historical wind speed data. The fitness value F(Li) is calculated based on the error between the predicted wind speed value and the actual wind speed value, and is defined as F(Li)=1 / (1+MSE(Li)), so that the larger the fitness value, the better the prediction effect of the weight combination;

[0066] For algorithms that also include selection operations, such as genetic algorithms, individuals in the current population are selected based on fitness values. Common selection algorithms include roulette selection and tournament selection. Through selection operations, individuals with excellent weighted combinations are retained.

[0067] Determine the best individual in the group, such as the leader goose of a flock of geese or the leading individual of a black-winged kite. Its position represents the optimal solution for the current weight. Then, according to the algorithm-specific update rules, let other individuals approach the best individual or explore new positions around it. The updated individual positions must be processed at the boundary to ensure that each weight value is still within the range of [wmin, wmax];

[0068] If it is a genetic algorithm, it is also necessary to perform a crossover operation on the selected individuals, select two individuals (parents) with probability Pc, exchange some genes (weight values) in their chromosomes (weight combinations), and generate new individuals (offspring). In addition, the offspring can also be mutated to randomly change some gene values ​​(weight values) in the chromosomes to enhance the diversity of the population and avoid falling into the local optimum.

[0069] Check whether the termination conditions are met. Common termination conditions include reaching the maximum number of iterations Tmax, the convergence of population fitness (that is, the change in fitness value for several consecutive generations is less than the set threshold), or finding a weight combination that meets specific accuracy requirements. If the termination conditions are met, the algorithm iteration is stopped, and the weight combination Li corresponding to the optimal individual of the current population is output as the optimal result. If not, continue iterative optimization.

[0070] At the same time, the kernel ridge regression model is introduced to determine the relationship between the prediction error and the predicted value, and the final wind speed prediction model is constructed:

[0071] Collect the predicted values ​​of the Vc group prediction model as A1p, A2p, ..., AVcp and the corresponding error dp to build a data set Among them, g is the total amount of predicted data;

[0072] By introducing the kernel function, the low-dimensional prediction value is converted into a high-dimensional feature representation to enhance the nonlinear fitting ability of the model. The Gaussian kernel function is selected here. According to the formula Map the input data to a high-dimensional feature space and calculate the kernel matrix K, which is positive definite and has dimension g×g, where xp=(A1 p, A2p, ..., AVcp), xq=(A1 q, A2q, ..., AVcq), and σ is the kernel bandwidth parameter;

[0073] Let the regression coefficient vector be a=(a0,a1,...,aVc) T ,Depend on As the minimization objective function, where λ is the regularization parameter;

[0074] Taking the partial derivative of J(a) with respect to a and setting it to 0, we get the linear variance group:

[0075] Where Q is a g-dimensional all-one vector, E is a g×g identity matrix, and d=(d1,d2,...,dg) T , solve the equation to obtain the regression coefficient a;

[0076] Based on the regression coefficient a, the functional relationship f between the prediction error and the predicted value is constructed:

[0077]

[0078] The final wind speed prediction model is w1A1+w2A2+...+wVcAVc+f(A1,A2,...,AVc).

[0079] The energy storage control module continuously monitors the wind speed status through the wind speed prediction module and calculates the wind speed change at adjacent moments k = (v(t2) - v(t1)) / (t2 - t1). If |k|>d, it means that the wind speed fluctuates violently at this time, and the energy storage device needs to be activated to regulate the balance of power transmission between the wind turbine and the power grid. If |k|<=d, it means that the wind speed changes smoothly at this time and the energy storage device does not need to be activated. Among them, v(t1) and v(t2) are the wind speeds at t1 and t2, and d is the wind speed change threshold.

[0080] The fan angle of attack adaptive control module, if |k|>d and k>0, adjusts the fan blade angle of attack according to the adaptive angle of attack increment:

[0081] Assume the maximum blade angle of attack is α max , which is usually determined by the aerodynamic design, material strength and overall mechanical structure tolerance of the wind turbine blades. For most common horizontal axis wind turbines, this value is generally in the range of 20-25 degrees;

[0082] Under the adaptive control strategy, the angle of attack increment Δα is not only related to the wind speed change k, but also needs to consider the current angle of attack α and the maximum value α max The specific adaptive angle of attack increment satisfies:

[0083] m is the proportional coefficient fitted according to the aerodynamic characteristics of the wind turbine and the power generation efficiency curve;

[0084] For example, a wind turbine has a 2.5MW wind turbine and a maximum angle of attack α max =22°, the current blade angle of attack ɑ=8°, m=3, the wind speed change k=2, the angle of attack increment calculated by the formula is △ɑ=3×2=6°, then the new angle of attack is ɑ+△ɑ=14°, which is within the range of the maximum angle of attack. Therefore, the angle of attack can be normally increased by 6°, and the wind turbine impeller can capture wind energy more efficiently to improve power generation efficiency and meet the performance optimization requirements of the current wind speed increase condition; on the contrary, in a certain extreme weather, the wind turbine is affected by wind speed fluctuations, the current angle of attack has reached 20°, the wind speed changes suddenly, k=1.5, and △ɑ=3×1.5=4.5° is calculated. If formula 1 is used at this time, the angle of attack will exceed the maximum angle of attack, so the actual △ɑ=22-20=2°, and the angle of attack is only increased by 2°, avoiding the risk of stalling and structural damage of the wind turbine due to excessive increase in the angle of attack, ensuring that the wind turbine is dynamically adjusted within the boundary of safe and efficient operation in the scene of rapid wind speed changes, and maintaining a balance between power generation stability and equipment reliability;

[0085] With the surge in power generation, in order to maintain the stability of the power grid, the excess power needs to be properly placed. The energy storage device accurately absorbs the excess power based on the real-time power difference. Assuming that the wind turbine power generation at time t is P(t), the grid can accept a stable power of Pg(t), and the energy storage device charging power Ps(t) follows:

[0086] Ensure that the power injected into the grid is stable and avoid grid frequency and voltage instability caused by a sudden increase in wind power;

[0087] The fan speed adaptive control module, if |k|>d and k<0, adjusts the fan speed according to the adaptive speed adjustment rate:

[0088] When the wind speed decreases rapidly, setting the adaptive speed adjustment rate can protect the mechanical components of the wind turbine while taking into account the power generation efficiency. The key here is to dynamically generate a reasonable speed adjustment instruction based on the real-time status of the wind turbine and the wind speed change to ensure that the wind turbine smoothly transitions through the sudden wind speed period.

[0089] Assume that the maximum allowable speed change rate that the mechanical components of the fan can withstand is Rmax, the current fan speed is N(t), and the constant based on the mechanical inertia of the fan and the performance of the speed control system is n. The specific adaptive speed adjustment rate satisfies:

[0090]

[0091] When the speed adjustment rate does not exceed Rmax, and the adjusted speed will not drop to a negative number, the speed is adjusted at a rate that matches the wind speed change, so that the fan can closely follow the wind speed and flexibly reduce speed;

[0092] When the speed adjustment rate exceeds Rmax, but the speed after adjustment is still not negative, in order to protect the equipment components, the adjustment rate is forcibly set to Rmax to ensure that there is no mechanical damage due to excessive speed reduction shock;

[0093] When the speed becomes negative, it means that the wind speed has dropped too much and may cause the fan to stop naturally. At this time, the current speed should be maintained and the wind speed should be waited for to rise or further instructions to be issued to avoid abnormal braking of the fan.

[0094] At the same time, the energy storage device discharge mode is turned on to fill the instantaneous power gap of the power grid and maintain the balance of power supply and demand. The discharge power Pd(t) is dynamically adjusted according to the real-time power shortage of the power grid. The real-time power demand of the power grid is Pr(t). The energy storage device discharge power Pd(t) follows:

[0095] It ensures continuous and stable power supply to key loads of the power grid and can always ensure the safety and efficiency of wind power equipment.

[0096] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0097] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy, characterized in that: include: In the wind speed prediction module, the KDBSCAN density clustering algorithm is used to divide the wind turbine group into n clusters. Each cluster has Vc (c∈[1,n]) similar wind turbines. HOT-RNN is used to train each wind turbine in the n clusters, and the trained Vc similar wind turbines in each cluster are combined. The corresponding weights are obtained through the optimization algorithm. At the same time, the kernel ridge regression model is introduced to determine the relationship between the prediction error and the prediction value, and the final wind speed prediction model is constructed. The energy storage control module continuously monitors the wind speed state through the wind speed prediction module, calculates the wind speed change k at adjacent moments, and activates the energy storage device if |k|>d, and does not activate the energy storage device if |k|<=d, where d is the wind speed change threshold; Wind turbine attack angle adaptive control module, if |k|>d and k>0, adjust the wind turbine blade attack angle according to the adaptive attack angle increment, if P(t)>Pg(t), activate the energy storage device for charging, where P(t) is the wind turbine power generation at time t, and Pg(t) is the stable power that the power grid can accept at time t; The wind turbine speed adaptive control module, if |k|>d and k<0, adjusts the wind turbine speed according to the adaptive speed adjustment rate. If Pr(t)>P(t), activates the energy storage device for discharge, where Pr(t) is the real-time power demand of the power grid.

2. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: The specific steps of using KDBSCAN density clustering algorithm to divide wind turbines include: Collect multi-dimensional data of wind turbines, including wind speed v, power p, longitude lon, latitude lat, generator speed g, and construct a data set X = {x1, x2, ..., xm}, xi = (vi, pi, loni, lati, gi), where m is the number of wind turbines in the wind turbine; Perform Z-score standardization on each xi in the data set. If there are missing values, use the mean to fill them. Determine the core point neighborhood radius eps according to the scale and distribution density of wind turbines, and set the minimum number of points min_samples; According to the formula Calculate the number of points in the eps neighborhood of point xi. If Neps(xi)>=min_samples, then xi is a core point, where d(xi,xj) is the Euclidean distance between two points, and I(·) is the indicator function, which is 1 if satisfied, otherwise 0; Starting from the core point xi, if xj is in the eps neighborhood of xi, xj and xi are grouped into one cluster; According to the formula Calculate the silhouette coefficient S to evaluate the clustering effect, and adjust the parameters accordingly to optimize the clustering, where ai is the average distance from xi to other points in the same cluster, and bi is the average distance from xi to all points in the nearest neighbor cluster.

3. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: The specific method of using HOT-RNN to train the fan is: Collect a large amount of historical operating data from the wind turbine monitoring system, including wind speed, wind direction, power output, humidity, ambient temperature, and air pressure, to construct a multivariate raw data set; Perform Min-Max normalization and data smoothing operations on the original data; The preprocessed data is divided into a training set and a test set at 80% and 20%, and the divided multivariate time series data is converted into a high-order tensor form; Determine the network architecture parameters, including the number of hidden layers, the number of neurons in each layer, the tensor order, and the time step; Based on the training set data, the mean square error (MSE) and mean absolute error (MAE) are used to measure the deviation between the predicted value and the true value, and Adam is used to iteratively update the model parameters; Substitute the test set data into the trained prediction model and calculate the MSE, MAE, and prediction accuracy to comprehensively measure the prediction accuracy, stability, and reliability of the model. If the requirements are not met, return to the above steps to continue training the model.

4. According to the wind turbine power grid management system based on wind speed prediction and adaptive control strategy described in claim 1, the optimization algorithm includes goose flock optimization algorithm, black kite optimization algorithm, genetic optimization algorithm, simulated annealing optimization algorithm, and particle swarm optimization algorithm.

5. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: The specific steps of introducing the kernel ridge regression model to determine the relationship between prediction error and prediction value are: Collect the prediction values ​​of the Vc group prediction model as A1 p, A2p, ..., AVcp and the corresponding error dp to build a data set Among them, g is the total amount of predicted data; According to the formula Map the input data to a high-dimensional feature space and calculate the kernel matrix K, where xp = (A1 p, A2p, ..., AVcp), xq = (A1 q, A2q, ..., AVcq), and σ is the kernel bandwidth parameter; Let the regression coefficient vector be a=(a0,a1,...,aVc) T ,Depend on As the minimization objective function, where λ is the regularization parameter; Take the partial derivative of J(a) with respect to a and set it to 0 to determine the regression coefficient a; Based on the regression coefficient a, the functional relationship f between the prediction error and the predicted value is constructed:

6. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: The final wind speed prediction model constructed is w1A1+w2A2+...+wVcAVc+f(A1,A2,...,AVc), where w1,w2,...,wVc are the weights calculated by the optimization algorithm.

7. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: According to the formula Set the adaptive angle of attack increment, where ɑ is the current angle of attack, max is the maximum blade angle of attack, and m is the proportional coefficient.

8. According to the wind turbine power grid management system based on wind speed prediction and adaptive control strategy as described in claim 7, according to the formula To adjust the energy storage device for charging, Ps(t) is the charging power of the energy storage device.

9. According to the wind turbine power grid management system based on wind speed prediction and adaptive control strategy as described in claim 1, according to the formula Get the fan adaptive speed adjustment rate, where: R max is the maximum allowable speed change rate of the fan, v is the current wind speed, N(t) is the current fan speed, and n1 is a constant.

10. A wind turbine power grid management system based on wind speed prediction and adaptive control strategy according to claim 1, characterized in that: According to the formula To adjust the energy storage device to discharge, where Pd(t) is the discharge power of the energy storage device.

Images