A wind moment matrix system power regulation method and device

Abstract

The application provides a wind matrix system power regulation method and device. The method comprises the following steps: obtaining an actual wind speed of a wind matrix system; inputting the actual wind speed into a pre-trained wind matrix power prediction model to predict a fan power of the wind matrix system; and inputting the fan power into a fuzzy self-adaptive PID controller to perform feedback adjustment on the fan power based on the fuzzy self-adaptive PID controller to obtain a target fan power. The method provided by the application can improve the wind field simulation accuracy by inputting the actual wind speed of the wind matrix system into the pre-trained wind matrix power prediction model to perform power prediction, and then inputting the predicted fan power into the fuzzy self-adaptive PID controller to perform feedback adjustment, so that the accuracy of the adjusted target fan power is higher, thereby improving the wind field simulation accuracy and providing more realistic and complex test conditions for unmanned aerial vehicle experiments.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind matrix system control, and particularly relates to a wind matrix system power regulation method and device. BACKGROUND

[0002] In the unmanned aerial vehicle experimental environment, the importance of the wind matrix cannot be ignored. As a key tool for simulating complex wind conditions in nature, it provides an ideal experimental platform for performance testing, structural verification, and optimization of flight control algorithms of unmanned aerial vehicles. Especially in the performance testing of unmanned aerial vehicles, the wind matrix can accurately create various complex and changeable airflow environments in the test site through the coordinated operation of multiple fans.

[0003] The flexibility of the wind matrix is one of its great advantages. By precisely regulating the working state of each fan, it can simulate airflow of different directions, intensities, and modes, enabling unmanned aerial vehicles to be tested in an environment close to real flight conditions. This highly simulated test environment is crucial for a comprehensive evaluation of the performance of unmanned aerial vehicles. For example, when simulating mountainous turbulent environments, the wind matrix can randomly change the direction and intensity of the airflow to test the flight stability and attitude control capability of unmanned aerial vehicles under complex airflow conditions. Such testing not only reveals the performance of unmanned aerial vehicles under extreme conditions, but also provides valuable experimental data for the optimization of flight control algorithms.

[0004] However, despite the significant achievements of the wind matrix in simulating various wind conditions, its simulation capabilities still need to be further improved when facing extreme or fine wind field characteristics. In particular, when it comes to complex wind field characteristics such as micro-scale turbulent structures, the simulation accuracy of the wind matrix often fails to meet ideal standards. This is mainly due to the insufficient power regulation accuracy of the wind matrix system, which makes it difficult to achieve the desired accuracy when simulating these fine wind field characteristics. SUMMARY

[0005] The present application provides a wind matrix system power regulation method and device to solve the defect of low power regulation accuracy of the wind matrix system in the prior art.

[0006] A wind matrix system power regulation method, comprising:

[0007] Obtaining the actual wind speed of the wind matrix system;

[0008] Inputting the actual wind speed into a pre-trained wind matrix power prediction model, and based on the model, predicting the fan power of the wind matrix system;

[0009] Inputting the fan power into a fuzzy self-adaptive PID controller, and based on the feedback adjustment of the fuzzy self-adaptive PID controller to the fan power, obtaining the target fan power of the wind matrix system.

[0010] Further, the wind matrix system power regulation method as described above, the pre-trained wind matrix power prediction model is trained by the following method:

[0011] Obtain the wind speed corresponding to each wind condition and the fan power corresponding to each wind speed of the wind matrix system;

[0012] Take the wind speed and fan power under all different wind conditions as training samples, train the initial wind matrix power prediction model, and obtain the trained wind matrix power prediction model.

[0013] Further, the wind matrix system power regulation method as described above, the initial wind matrix power prediction model is constructed by using a CNN model.

[0014] Further, the wind matrix system power regulation method as described above, the parameters of the fuzzy adaptive PID controller are the optimal parameters obtained based on the simulated annealing algorithm.

[0015] Further, the wind matrix system power regulation method as described above, the parameters of the fuzzy adaptive PID controller are the optimal parameters obtained based on the simulated annealing algorithm, which includes:

[0016] Obtain the initial parameters of the fuzzy adaptive PID controller;

[0017] Randomly perturb the initial parameters at the current temperature to obtain updated parameters;

[0018] Run the wind matrix system based on the fuzzy adaptive PID controller corresponding to the updated parameters, obtain the stable wind speed of the wind matrix system within a preset time and the power corresponding to the stable wind speed;

[0019] Determine the target function value corresponding to the current temperature according to the stable wind speed and the power corresponding to the stable wind speed;

[0020] Compare the target function value J corresponding to the current temperature with the initial minimum target function value J' min , and determine the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value J min according to the comparison result.

[0021] Iteratively update the current temperature, and update the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value J min based on the target function value corresponding to the updated temperature, until the updated temperature is lower than a preset termination temperature, and then output the final optimal parameters of the fuzzy adaptive PID controller.

[0022] Further, the wind matrix system power regulation method as described above, the target function value J corresponding to the current temperature is compared with the initial minimum target function value J' min , and the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value J min are determined according to the comparison result.

[0023] In the case that the target function value J corresponding to the current temperature is less than the initial minimum target function value J' min , the updated parameters are taken as the current optimal parameters of the fuzzy adaptive PID controller, and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value J mim .

[0024] Further, the wind matrix system power regulation method as described above, in the case that the target function value J corresponding to the current temperature is greater than or equal to the initial minimum target function value J' min , the acceptance probability P is calculated, and the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value J min are determined based on the acceptance probability P.

[0025]

[0026] Wherein, J` is the target function value corresponding to the current temperature, J min is the current minimum target function value, and T is the current temperature.

[0027] Further, the wind matrix system power regulation method as described above, the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value J mim are determined based on the acceptance probability P.

[0028] In the case that the acceptance probability P is greater than the preset random parameter p, the updated parameters are taken as the current optimal parameters of the fuzzy adaptive PID controller, and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value J min .

[0029] In the case that the acceptance probability P is less than or equal to the preset random parameter p, the optimal parameters corresponding to the last updated temperature are taken as the current optimal parameters of the fuzzy adaptive PID controller, and the target function value J corresponding to the last updated temperature is taken as the current minimum target function value J min .

[0030] Further, the wind matrix system power regulation method as described above, the obtaining the initial parameters of the fuzzy adaptive PID controller comprises:

[0031] According to experience, set the initial fuzzy adaptive PID control parameters of the PID controller;

[0032] Based on the initial fuzzy adaptive PID control parameters corresponding to the fuzzy adaptive PID controller, the wind matrix system is run to determine the current actual wind speed of the wind matrix system;

[0033] According to the current actual wind speed and the target wind speed of the wind matrix system, the error e and the error change rate ec of the wind speed are determined;

[0034] According to the pre-defined fuzzy subsets and membership functions, the error e and the error change rate ec of the wind speed are fuzzified into corresponding fuzzy values;

[0035] According to the fuzzy control rule table, the parameter adjustment amount of the fuzzy adaptive PID controller is determined by fuzzy reasoning;

[0036] Based on the barycentric method, the parameter adjustment amount is converted into the accurate parameters of the fuzzy adaptive PID controller, and the accurate parameters are used as the initial parameters of the fuzzy adaptive PID controller.

[0037] The application also provides a wind matrix system power regulation device, comprising:

[0038] An obtaining unit is configured to obtain the actual wind speed of the wind matrix system;

[0039] A prediction unit is configured to input the actual wind speed into a pre-trained wind matrix power prediction model, and predict the fan power of the wind matrix system based on the model;

[0040] An adjusting unit is configured to input the fan power into a fuzzy adaptive PID controller, and adjust the fan power based on the fuzzy adaptive PID controller to obtain the target fan power of the wind matrix system.

[0041] The wind matrix system power regulation method and device provided by the application can improve the wind field simulation accuracy by inputting the actual wind speed of the wind matrix system into a pre-trained wind matrix power prediction model for power prediction, and then adjusting the predicted fan power through a fuzzy adaptive PID controller, so as to obtain a target fan power with higher accuracy, thereby providing a test condition closer to a real complex environment for unmanned aerial vehicle experiments. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to make the technical solutions in the present application or the prior art clearer, the drawings needed in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0043] Figure 1 The flow chart of the wind matrix system power regulation method provided by the present application;

[0044] Figure 2 The schematic diagram of the wind matrix system structure provided by the present application;

[0045] Figure 3 The flow chart of the simulated annealing algorithm provided by the present application;

[0046] Figure 4 The structure block diagram of the wind matrix system power regulation device provided by the present application. DETAILED DESCRIPTION

[0047] In order to make the technical solutions in the present application or the prior art clearer, the drawings needed in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0048] In the UAV experimental environment, the setting of the wind matrix is crucial. According to the specific needs of the experiment, the layout and scale of the wind matrix will be carefully designed. These wind matrices are composed of multiple fans and work collaboratively through advanced control systems, enabling the simulation of various airflow directions, intensities, and patterns. UAVs perform flight tests in such airflow environments, and experimenters can simulate different meteorological conditions by adjusting the parameters (wind speed, power) of the wind matrix, providing a comprehensive test environment for the performance evaluation of UAVs. Although the wind matrix can generate a variety of wind conditions, current technology still has deficiencies in simulating some extreme or fine wind field characteristics. For example, when simulating micro-scale turbulent structures, it may not be able to accurately reproduce the spectral characteristics and spatial distribution of turbulence in nature. This leads to some differences between the experimental environment and the actual complex environment, affecting the accuracy of the performance evaluation of UAVs under extreme conditions. In order to solve this problem, the present application applies neural networks and fuzzy adaptive PID controllers to the control of the wind matrix system.

[0049] Among them, the neural network, with its powerful nonlinear mapping ability and self-learning ability, has shown excellent performance in dealing with complex and uncertain system problems. In the control of the wind matrix system, the neural network can learn the potential correlation between the fan power and the wind speed through the training of a large amount of historical data, thereby realizing the accurate prediction and control of the fan power. This data-based control method not only improves the accuracy of wind speed control, but also reduces the energy consumption of the system to a certain extent. On the other hand, the fuzzy adaptive PID controller combines the advantages of fuzzy logic and PID control, and can dynamically adjust the control parameters according to the real-time state of the system, so that the control system can quickly respond to external disturbances or internal changes of the system, and maintain stable control effect. In the wind matrix system, the fuzzy adaptive PID controller can flexibly adjust the power output of the fan according to the real-time feedback of the wind speed, to ensure that the wind speed always maintains within the set target range. The invention combines neural networks with fuzzy adaptive PID controllers to further improve the control performance of the wind matrix system. Specifically, the neural network can be used to build a prediction model between the fan power and the wind speed, providing accurate reference information for the fuzzy adaptive PID controller; and the fuzzy adaptive PID controller can adjust the control strategy in real time according to these prediction information, to ensure the accuracy and stability of the wind speed control. The scheme provided by the invention is described in detail as follows:

[0050] Figure 1 The power regulation method flow chart of the wind matrix system provided by the invention is shown in Figure 1 The method comprises the following steps:

[0051] Step 1: Obtain the actual wind speed of the wind matrix system.

[0052] Specifically, Figure 2 The structure diagram of the wind matrix system provided by the invention is shown in Figure 2 The actual wind speed of the wind matrix is collected by the wind speed sensor. The control system includes a fuzzy adaptive PID controller and a power prediction model; the actual wind speed collected by the wind speed sensor is input into the power prediction model, and the fan power corresponding to the actual wind speed is predicted based on the power prediction model; the predicted fan power is input into the fuzzy adaptive PID controller, and the adjusted fan power is finally output through the feedback adjustment of the fuzzy adaptive PID controller; and the industrial control system controls the operation of the wind matrix using the fan power, so as to simulate the required meteorological conditions of the wind matrix.

[0053] In the embodiment of the invention, the wind matrix system is composed of 40x40 fan combinations, and the corresponding wind speed is accurately collected by using a 6x6 wind speed sensor array in front of the wind matrix to obtain a wind speed matrix.

[0054] Step two: input the actual wind speed into the pre-trained wind matrix power prediction model, and based on the model, the fan power of the wind matrix system is predicted.

[0055] Specifically, the actual wind speed matrix measured by the sensor is input into the pre-trained wind matrix power prediction model, the pre-trained wind matrix power prediction model extracts and analyzes the wind speed matrix, converts the wind speed matrix information into a preliminary estimate of the wind matrix power control signal, and outputs the power adjustment suggestion value corresponding to each fan.

[0056] Step three: input the fan power into the fuzzy adaptive PID controller, and based on the fuzzy adaptive PID controller, the fan power is adjusted to obtain the target fan power of the wind matrix system.

[0057] Specifically, the power adjustment suggestion value corresponding to each fan output by the wind matrix power prediction model is input into the fuzzy adaptive PID controller, and the power corresponding to each fan is adjusted through the fuzzy adaptive PID controller, and finally the adjusted target fan power is obtained.

[0058] The wind matrix system power regulation method provided by the application has the advantages that by inputting the actual wind speed of the wind matrix system into the pre-trained wind matrix power prediction model for power prediction, and then adjusting the predicted fan power through the fuzzy adaptive PID controller, the accuracy of the adjusted target fan power is higher, thereby improving the wind field simulation accuracy and providing more realistic test conditions for unmanned aerial vehicle experiments.

[0059] Further, the pre-trained wind matrix power prediction model is trained by the following method:

[0060] Firstly, the wind speed corresponding to each wind speed and the fan power corresponding to each wind speed of the wind matrix system under different wind conditions are obtained; secondly, the wind speed and fan power under different wind conditions are used as training samples to train the initial wind matrix power prediction model to obtain the trained wind matrix power prediction model.

[0061] Specifically, the training process of the model is as follows:

[0062] 1. Data collection:

[0063] After starting the wind matrix system, the data acquisition device starts to collect wind speed and fan power data, and the sampling frequency is 1 data per second. A large number of experiments are carried out in the wind matrix system, and the power combination of 40x40 fans is changed comprehensively, and the corresponding wind speed mode data is accurately collected by using the 6x6 wind speed sensor array in front of the wind matrix.

[0064] 2. Data set preprocessing:

[0065] Data normalization:

[0066] Preprocessing the data using mean filtering, and normalizing the wind speed data and fan power data respectively to the interval (0, 1). The specific formula is as follows:

[0067]

[0068] Where Z is the normalized data, X is the original data, X_min and X_max are the minimum and maximum values of the original data respectively. In order to better handle data of different orders of magnitude in model training.

[0069] For the wind speed matrix A, its size is M x N, and the window size is (m x n). The filtered matrix B has a size of (M-m+1) x (N-n+1), and its calculation formula is:

[0070]

[0071] 3. Data set segmentation:

[0072] The collected preprocessed data is divided into training set, validation set and test set according to the proportion of 70%, 20% and 10%.

[0073] 4. CNN model construction:

[0074] Construct a CNN model structure suitable for wind matrix power-wind speed relationship processing. After multiple experiments and model performance evaluation, it is determined that the combination of 3 convolution layers and 2 pooling layers can achieve a good balance between model complexity and performance.

[0075] Output layer: 6*6 matrix, input channel 1.

[0076] First convolution layer: convolution kernel size is set to 5x5 and step is 1, convolution kernel number is 16, input channel number is 1 (consistent with input layer), output channel number is 16.

[0077] First pooling layer: following the first convolution layer, its window size is 2x2 and step is 2. The input channel number of this pooling layer is 16 (the same as the output channel number of the last convolution layer), and the output channel number is also 16.

[0078] Second convolution layer: convolution kernel size is still 5x5 and step is 1, convolution kernel number increases to 32, input channel number is 16 (the same as the output channel number of the last pooling layer), and output channel number is 32.

[0079] The second pooling layer: the same 2*2 window size and 2 step size are adopted for the pooling operation, the input channel number is 32 (the same as the output channel number of the last convolution layer), and the output channel number is 32.

[0080] The third convolution layer: the convolution kernel size is kept as 5*5 and the step size is 1, the convolution kernel number is increased to 64, the input channel number is 32 (the same as the output channel number of the last pooling layer), and the output channel number is 64.

[0081] The full connection layer: the weight parameter matrix size of the full connection layer is 1024*1600, and the bias parameter vector size is 1600.

[0082] The activation function: the ReLU function is used as the activation function.

[0083] In the embodiment of the application, the initial wind matrix power prediction model is constructed by using a CNN model.

[0084] Specifically, in the wind matrix power prediction, the CNN model can effectively extract complex spatio-temporal features in the wind power time series through multiple convolution and pooling operations. This feature extraction capability enables the CNN model to more accurately capture the variation law of wind power, thereby improving the prediction accuracy of power. At the same time, the CNN model can also use cross-validation and other methods for parameter optimization, further improving the prediction performance of the model. This efficient model training and optimization capability enables the CNN model to have faster training speed and better generalization ability in wind matrix power prediction. Moreover, the CNN model has strong adaptability and scalability. It can adapt to different scales of wind farms and different wind power prediction scenarios. With the continuous accumulation of wind power data and the improvement of computing power, the CNN model can be further extended and optimized to cope with more complex wind power prediction problems.

[0085] Further, the parameters of the fuzzy self-adaptive PID controller provided by the application are the optimal parameters obtained based on the simulated annealing algorithm.

[0086] Specifically, the simulated annealing algorithm can effectively explore the global optimal solution in a huge solution space. When optimizing the parameters of the fuzzy self-adaptive PID controller, the parameters of the fuzzy self-adaptive PID controller can be gradually adjusted through the iteration process of the simulated annealing algorithm, thereby improving the accuracy of the parameters. The present application improves the existing simulated annealing algorithm, and optimizes the parameters of the fuzzy self-adaptive PID controller by using the improved simulated annealing algorithm. The optimization method is described in detail as follows, Figure 3 The flow chart of the simulated annealing algorithm provided by the application is shown in Figure 3 The method comprises the following steps:

[0087] Step 1: obtaining the initialization parameters of the fuzzy self-adaptive PID controller.

[0088] wherein the initialization parameters of the fuzzy adaptive PID controller are obtained according to the following method:

[0089] According to experience, the initial fuzzy adaptive PID control parameters of the PID controller are set; based on the fuzzy adaptive PID controller corresponding to the initial fuzzy adaptive PID control parameters, the wind turbine system is run to determine the current actual wind speed of the wind turbine system; according to the current actual wind speed and the target wind speed of the wind turbine system, the error e and the error change rate ec of the wind speed are determined; according to the pre-defined fuzzy subsets and membership functions, the error e and the error change rate ec of the wind speed are fuzzified into corresponding fuzzy values; according to the fuzzy control rule table, the parameter adjustment amount of the fuzzy adaptive PID controller is determined through fuzzy reasoning; based on the gravity center method, the parameter adjustment amount is converted into the accurate parameters of the fuzzy adaptive PID controller, and the accurate parameters are taken as the initialization parameters of the fuzzy adaptive PID controller.

[0090] Specifically, according to previous experience, the fuzzy adaptive PID controller parameters are initially set, the initial value of the proportional coefficient K p is set to 1.0, the initial value of the integral coefficient K i is set to 0.05, and the initial value of the differential coefficient K d is set to 0.5.

[0091] The fuzzy subsets {negative big (NB), negative medium (NM), zero (ZO), positive medium (PM), positive big (PB)} are defined.

[0092] The membership function adopts a triangular membership function,

[0093] The calculation formula is as follows:

[0094] When x < a, μ(x) = 0;

[0095] When a ≤ x < β,

[0096] When x = b, μ(x) = 1;

[0097] When b < x ≤ c,

[0098] When x > c, μ(x) = 0.

[0099] Wherein x is the input variable, a is the lower limit value, b is the peak value, and c is the upper limit value.

[0100] Fuzzy set and membership function parameter definition of error e and error change rate ec

[0101] Table 1:

[0102]

[0103] Table 2:

[0104]

[0105]

[0106] The fuzzy adaptive PID controller obtains the power control signal output by the CNN network and the error e and error change rate ec of the current actual wind speed and the target wind speed.

[0107] Fuzzy step: According to the pre-defined fuzzy subsets and membership functions, the error e and error change rate ec are fuzzified into corresponding fuzzy values. For example, the error e is fuzzified into negative big (NB), negative medium (NM), zero (ZO), positive medium (PM), positive big (PB), etc. fuzzy subsets, and its membership degrees in each fuzzy subset are determined.

[0108] Fuzzy reasoning step: fuzzy reasoning is performed according to the fuzzy control rule table. According to the combination of the fuzzy values of the error e and the error change rate ec, the fuzzy output of the adjustment amount ΔK p , ΔK i , ΔK d of the PID parameters is determined according to the rule table.

[0109] Defuzzification step: the barycentric method is used to convert the fuzzy output of the PID parameter adjustment amount into an accurate value. These accurate PID parameter adjustment amounts are added to the current K p , K i , K d , and the parameters of the PID controller are updated.

[0110] Finally, the PID controller adjusts the wind matrix power control signal output by the CNN network according to the updated parameters to obtain the final power control signal, which is sent to the power controller. The power controller adjusts the actual power output of the fan according to the signal.

[0111] Step 2: Randomly perturb the initialization parameters at the current temperature to obtain updated parameters (K` p , K` i , K` d ).

[0112] Step 3: Based on the fuzzy adaptive PID controller corresponding to the updated parameters, run the wind matrix system to obtain the stable wind speed of the wind matrix system within a predetermined time and the power corresponding to the stable wind speed.

[0113] Specifically, the updated parameters (K` p , K` i , K` d) applied to the fuzzy adaptive PID controller of the wind matrix system, run the wind matrix system for 10 minutes, collect the wind speed data and fan power data during this period. Among them, the wind speed data and fan power data are an average wind speed value in the stable interval within 10 minutes of running, and the power is the average power value corresponding to the stable interval of the wind speed.

[0114] Step 4: According to the stable wind speed and the power corresponding to the stable wind speed, determine the target function value corresponding to the current temperature.

[0115] Specifically, where the target function is J = w1*MSE + w2*Overshoot + w3*Energy, the embodiment of the application defines the target function in the form of weighted sum, wherein MSE is the mean square error, used to measure the accuracy of wind speed control, and the calculation formula is Where N is the sample size, y i is the set wind speed value, is the stable wind speed value; Overshoot is the overshoot; Energy is the power corresponding to the stable wind speed. w1, w2, w3 are weight coefficients.

[0116] Step 5: Compare the target function value J corresponding to the current temperature with the initial minimum target function value J' min .

[0117] Step 6: If the target function value J corresponding to the current temperature is less than the initial minimum target function value J' min , then update the parameters as the current optimal parameters of the fuzzy adaptive PID controller; and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value J min .

[0118] Step 7: If the target function value J corresponding to the current temperature is greater than or equal to the initial minimum target function value J' min , calculate the acceptance probability P;

[0119]

[0120] Where J` is the target function value corresponding to the current temperature, and J min is the current minimum target function value.

[0121] Step 8: Generate a random number p uniformly distributed in the interval [0, 1], and compare the acceptance probability P with the preset random parameter p.

[0122] Step 9: In the case that the acceptance probability P is greater than the preset random parameter p, the updated parameter is taken as the current optimal parameter of the fuzzy adaptive PID controller, and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value J min .

[0123] Step 10: In the case that the acceptance probability P is less than or equal to the preset random parameter p, the optimal parameter corresponding to the last updated temperature is taken as the current optimal parameter of the fuzzy adaptive PID controller, and the target function value J corresponding to the last updated temperature is taken as the current minimum target function value J min .

[0124] Specifically, a random number p uniformly distributed in the interval [0, 1] is generated. If P > p, the new parameter solution (K` P , K` i , K` d ) is accepted, that is, the current optimal parameter combination is updated to the new parameter combination, and J min = J` is updated; if P≤p, the new parameter solution is rejected, and the current optimal parameter combination is retained.

[0125] Step 11: The current temperature is iteratively updated to obtain an updated temperature.

[0126] Step 12: Based on the target function value corresponding to the updated temperature, the initial optimal parameter of the fuzzy adaptive PID controller and the current minimum target function value J mi are updated.

[0127] Step 13: Determine whether the updated temperature is lower than the preset termination temperature T min . If it is lower than the preset termination temperature T min , output the final optimal parameter corresponding to the fuzzy adaptive PID controller.

[0128] Specifically, the initial temperature T0 is set to 100℃, the temperature drop coefficient a is taken as 0.95℃, the termination temperature T min is set to 1, and the iteration number at each temperature is set to 100. When L iterations are completed at the current temperature T, the temperature is updated according to the temperature drop formula T = a*T. Then determine whether the current temperature T is lower than the termination temperature T min . If it is lower, stop the search process of the simulated annealing algorithm, and output the current optimal fuzzy adaptive PID parameter combination; if it is higher, return to the step of generating a new parameter solution and continue the parameter search at the new temperature.

[0129] The wind matrix system power regulation device provided by the present application is described below, and the wind matrix system power regulation device described below can be correspondingly referred to the wind matrix system power regulation method described above.

[0130] The present application also provides a wind matrix system power regulation device, Figure 4 The wind matrix system power regulation device provided by the present application is described below, and the wind matrix system power regulation device described below can be correspondingly referred to the wind matrix system power regulation method described above. Figure 4 As shown in the structural block diagram of the wind matrix system power regulation device provided by the present application,

[0131] The acquisition unit 401 is configured to acquire an actual wind speed of the wind matrix system.

[0132] The prediction unit 402 is configured to input the actual wind speed into a pre-trained wind matrix power prediction model, and predict a fan power of the wind matrix system based on the model.

[0133] The adjustment unit 403 is configured to input the fan power into a fuzzy self-adaptive PID controller, and perform feedback adjustment on the fan power based on the fuzzy self-adaptive PID controller to obtain a target fan power of the wind matrix system.

[0134] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for power regulation of a wind turbine system, characterized by, The method comprises the following steps: obtaining an actual wind speed of a wind matrix system; inputting the actual wind speed into a pre-trained wind matrix power prediction model, and predicting a fan power of the wind matrix system based on the model; inputting the fan power into a fuzzy self-adaptive PID controller, and adjusting the fan power based on the fuzzy self-adaptive PID controller to obtain a target fan power of the wind matrix system; the pre-trained wind matrix power prediction model is obtained by the following method: obtaining wind speeds corresponding to different wind conditions of the wind matrix system and fan powers corresponding to each wind speed; using all the wind speeds and fan powers under different wind conditions as training samples to train an initial wind matrix power prediction model to obtain the trained wind matrix power prediction model; the parameters of the fuzzy self-adaptive PID controller are optimal parameters obtained based on a simulated annealing algorithm; the parameters of the fuzzy self-adaptive PID controller are optimal parameters obtained based on a simulated annealing algorithm, which comprises the following steps: obtaining initialization parameters of the fuzzy self-adaptive PID controller; randomly perturbing the initialization parameters at a current temperature to obtain updated parameters; running the wind matrix system based on the fuzzy self-adaptive PID controller corresponding to the updated parameters, obtaining a stable wind speed of the wind matrix system within a preset time and a power corresponding to the stable wind speed; determining a target function value corresponding to the current temperature according to the stable wind speed and the power corresponding to the stable wind speed; comparing the target function value J corresponding to the current temperature with an initial minimum target function value determining the initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value according to a comparison result ; iteratively update the current temperature, and based on a target function value corresponding to the updated temperature, update the initial optimal parameter of the fuzzy adaptive PID controller and the current minimum target function value update until the updated temperature is lower than a preset termination temperature, and output a final optimal parameter corresponding to the fuzzy adaptive PID controller.

2. The wind matrix system power regulation method of claim 1, wherein, the initial wind matrix power prediction model is constructed by using a CNN model.

3. The wind matrix system power regulation method of claim 1, wherein, The target function value J corresponding to the current temperature is compared with an initial minimum target function value The initial optimal parameters of the fuzzy adaptive PID controller and the current minimum target function value are determined according to a comparison result The method comprises the steps of: In the case where the target function value J corresponding to the current temperature is less than the initial minimum target function value , the updated parameter is taken as the current optimal parameter of the fuzzy adaptive PID controller, and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value .

4. The wind matrix system power regulation method of claim 3, wherein, In the case where the target function value J corresponding to the current temperature is greater than or equal to the initial minimum target function value , a probability of acceptance P is calculated, and based on the probability of acceptance P, an initial optimal parameter of a fuzzy adaptive PID controller and a current minimum target function value are determined . wherein, is the target function value corresponding to the current temperature, is the current minimum target function value, and T is the current temperature.

5. The wind matrix system power regulation method of claim 4, wherein, The initial optimal parameters of the fuzzy adaptive PID controller and the current minimum objective function value are determined based on the acceptance probability P comprising: In the case that the acceptance probability P is greater than a preset random parameter , the update parameter is taken as the current optimal parameter of the fuzzy adaptive PID controller, and the target function value J corresponding to the current temperature is taken as the updated current minimum target function value . In the case that the acceptance probability P is less than or equal to the preset random parameter , the optimal parameter corresponding to the last updated temperature is taken as the current optimal parameter of the fuzzy adaptive PID controller, and the target function value J corresponding to the last updated temperature is taken as the current minimum target function value .

6. The wind matrix system power regulation method of claim 1, wherein, The initialization parameters of the fuzzy self-adaptive PID controller are obtained by the following steps: The initial fuzzy adaptive PID control parameters of the PID controller are set, the initial value of the proportional coefficient is set as 1.0, the initial value of the integral coefficient is set as 0.05, and the initial value of the differential coefficient is set as 0.5 ;​​ running the wind matrix system based on the fuzzy self-adaptive PID controller corresponding to the initial fuzzy self-adaptive PID control parameters to determine a current actual wind speed of the wind matrix system; determining an error in wind speed based on the current actual wind speed and a target wind speed of the wind matrix system and a rate of change of error ; According to a predefined fuzzy subset and membership function, the error of the wind speed and the error change rate is fuzzified into a corresponding fuzzy value; determining a parameter adjustment amount of the fuzzy self-adaptive PID controller according to a fuzzy control rule table and fuzzy reasoning; converting the parameter adjustment amount into accurate parameters of the fuzzy self-adaptive PID controller based on a gravity center method, and using the accurate parameters as the initialization parameters of the fuzzy self-adaptive PID controller.

7. An apparatus for implementing the power regulation method of a wind matrix system as described in claim 1, characterized in that, The method comprises the following steps: an obtaining unit is configured to obtain an actual wind speed of a wind matrix system; a prediction unit is configured to input the actual wind speed into a pre-trained wind matrix power prediction model, and predict a fan power of the wind matrix system based on the model; an adjusting unit is configured to input the fan power into a fuzzy self-adaptive PID controller, and adjust the fan power based on the fuzzy self-adaptive PID controller to obtain a target fan power of the wind matrix system.

Images