Global Optimal Parameters Identification Method for Direct-Drive Permanent Magnet Synchronous Wind Generator

Through particle swarm optimization algorithm and noise covariance matrix update, combined with time and measurement update, the problem of easy impact on the parameters of direct-drive permanent magnet synchronous wind turbine is solved, achieving high-precision and fast global optimal parameter identification, improving the The accuracy and response speed of the system are achieved.

CN113098341BInactive Publication Date: 2024-11-26HUNAN UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202110366185.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-04-06
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The parameters of direct-drive permanent magnet synchronous wind turbines are easily affected by unpredictable factors, resulting in poor vector control effect, and the controller parameters need to be frequently adjusted to achieve high-precision control. The existing methods are prone to local optimality and noise covariance matrix. It is difficult to select, affecting the accuracy and speed of parameter identification.

Method used

The particle swarm optimization algorithm is used to initialize the system noise and measurement noise covariance matrix, and the global optimal estimation of the system state is achieved through iterative updates and fitness function optimization, combining time and measurement updates, and adding average position and extreme perturbations.

Benefits of technology

The parameter identification accuracy and response speed of direct-drive permanent magnet synchronous wind turbine are improved, and local optimization is avoided, noise covariance matrix is ​​optimized, and high-precision and rapid identification is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113098341B_ABST
    Figure CN113098341B_ABST
Patent Text Reader

Abstract

The present invention discloses a global optimal parameter identification method for a direct-drive permanent magnet synchronous wind turbine, the steps of which include: initializing a particle group, giving the position and velocity initial values ​​of the system noise covariance matrix Q and the measurement noise covariance matrix R; finding the best solution to the problem through continuous iteration, updating the Q value and the R value; updating the current individual average extreme value and the group extreme value by the fitness function; according to the number of evolutionary stagnation of the individual average extreme value and the group extreme value, and perturbing the extreme value; obtaining the optimal identification value of the system state at that moment through measurement update and time update. The present invention can realize high-precision identification of the stator resistance, dq inductance and permanent magnet flux of the direct-drive permanent magnet synchronous wind turbine, and has the advantages of fast speed and strong generalization ability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of direct-drive permanent magnet synchronous wind generators, and in particular relates to a global optimal parameter identification method for a direct-drive permanent magnet synchronous wind generator. Background Art

[0002] The direct-drive permanent magnet synchronous wind turbine is a strongly coupled, multivariable, nonlinear system. The parameters of the generator are easily affected by unpredictable factors. For example, the resistance of the generator is greatly affected by temperature, the AC and DC axis inductance current, and the permanent magnet flux linkage are greatly affected by flux saturation, which makes the vector control unable to achieve the ideal effect. In addition, the parameter setting of the controller is related to the generator parameters. When the generator parameters change, the parameters of the controller need to be re-set, otherwise the high-precision control requirements cannot be met. Therefore, it is necessary to design a global optimal parameter identification method for direct-drive permanent magnet synchronous wind turbines with high accuracy, fast speed, and strong generalization ability. Summary of the invention

[0003] In view of the problems described in the background technology, the present invention provides a global optimal parameter identification method for a direct-drive permanent magnet synchronous wind turbine generator, which improves the accuracy, response speed and generalization ability of the system.

[0004] Global optimal parameter identification method for direct-drive permanent magnet synchronous wind turbine generator,

[0005] The steps include:

[0006] (1) Initialize the particle swarm, and give the initial values ​​of the position and velocity of the system noise covariance matrix Q and the measurement noise covariance matrix R;

[0007] (2) Passing

[0008]

[0009] Continuously iterate to find the best solution to the problem, update the Q value and R value; where ν ij is the velocity vector of particle i in the jth dimension; p ij is the position vector of particle i in the jth dimension, P best-md is the current best individual average extreme value of particle i in the space search; G best-d is the current best group extreme value searched in the population; k represents the current number of iterations; c1, c2 are the state vectors of the particles; r1, r2 are random numbers between 0 and 1; w is the weight factor;

[0010] (3) The current best individual average extreme value and the current best group extreme value are updated by the fitness function;

[0011] (4) According to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value, the extreme value is disturbed;

[0012] (5) Obtain the optimal estimate of the system state at the current moment through time update and measurement update.

[0013] 2. The system noise covariance matrix Q and the measurement noise covariance matrix R can be expressed as:

[0014]

[0015] Where W is the process noise vector; V is the observation noise vector; E{} represents the mathematical expectation.

[0016] 3. The step of updating the current best individual average extreme value and the current best group extreme value by the fitness function includes:

[0017] 1) According to the formula

[0018]

[0019] Calculate the fitness function; where M is the total number of iterations; i d is the stator d-axis current detection value; i q is the stator q-axis current detection value; is the estimated value of stator d-axis current; is the estimated value of stator d-axis current;

[0020] 2) If the fitness function is 0, update the current best individual average extreme value and the current best group extreme value, otherwise, keep the original parameter value.

[0021] 4. The method of perturbing the extreme value according to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value is as follows:

[0022]

[0023] in:

[0024]

[0025] n represents the number of evolutionary stagnation times of the current best individual extreme value and the current best group extreme value, and N represents the stagnation step threshold that the current best individual extreme value and the current best group extreme value need to be disturbed.

[0026] 5. The step of obtaining the optimal estimated value of the system state at the current moment through time updating and measurement updating comprises:

[0027] [1] Establish discrete state equations and measurement equations

[0028]

[0029] Observation vector for flux linkage and inductance identification Resistance Identification Observation Vector

[0030] [2] Expand f[x(k),u(k)] and h[x(k+1)] in a first-order Taylor series;

[0031]

[0032] in: x(k) is the state vector of the system; φ(k) is the state transfer matrix; B(k) is the control input matrix; u(k) is the input vector of the system; z(k+1) is the observation vector of the system; H(k) is the state observation matrix; w(k) is the process noise; v(k+1) is the observation noise; T S is the sampling time;

[0033] [3] Time update includes predicting the state at the next moment and the prediction error covariance matrix based on the input and the last estimated value. The method for predicting the state at the next moment based on the input and the last estimated value is:

[0034]

[0035] Among them, the superscript "^" represents the state estimation value, the superscript "~" represents the predicted value; "(k|k)" represents the initial value; "(k+1|k)" represents the value of the k+1 step predicted by the kth step; and "(k+1|k+1)" represents the output value after the iteration ends.

[0036] The method of the prediction error covariance matrix is

[0037]

[0038] [4] The measurement update includes calculating the Kalman gain and updating the state variables to be identified by the system's observation vector: The method for calculating the Kalman gain is:

[0039]

[0040] The method of updating the state variables to be identified by the system's observation vector is:

[0041]

[0042] The benefits of the present invention compared with the prior art are:

[0043] (1) This method solves the problem that traditional methods are prone to falling into local optimality. By adding the best average position and extreme value perturbations to optimize the identification parameters, the global optimality of the algorithm is achieved.

[0044] (2) In order to solve the problem of difficulty in selecting the system and measurement noise covariance matrix in traditional methods, this method optimizes and updates the error covariance matrix through time update and measurement update, thereby achieving high-precision and rapid identification of direct-drive permanent magnet synchronous wind turbine parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a basic flow chart of the implementation method of the present invention. DETAILED DESCRIPTION

[0046] In order to describe the present invention more specifically, the present invention is further described below in conjunction with the accompanying drawings and implementation modes.

[0047] like Figure 1 As shown, the global optimal parameter identification method of the direct-drive permanent magnet synchronous wind turbine generator comprises the following steps:

[0048] (1) Initialize the particle swarm, and give the initial values ​​of the position and velocity of the system noise covariance matrix Q and the measurement noise covariance matrix R;

[0049] (2) Use equation (1) to continuously iterate and find the best solution to the problem, and update the Q value and R value;

[0050]

[0051] In the above formula, ν ij is the velocity vector of particle i in the jth dimension; p ij is the position vector of particle i in the jth dimension, P best-md is the current best individual average extreme value of particle i in the space search; G best-d is the current best group extreme value searched in the population; k represents the current number of iterations; c1, c2 are the state vectors of the particles; r1, r2 are random numbers between 0 and 1; w is the weight factor.

[0052] (3) The current best individual average extreme value and the current best group extreme value are updated by the fitness function;

[0053] (4) According to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value, the extreme value is disturbed;

[0054] (5) Obtain the optimal estimate of the system state at that moment through time update and measurement update.

[0055] The system noise covariance matrix Q and the measurement noise covariance matrix R can be expressed as:

[0056]

[0057] In the above formula, W is the process noise vector; V is the observation noise vector; E{} represents the mathematical expectation.

[0058] The steps of updating the current best individual average extreme value and the current best group extreme value by the fitness function include:

[0059] 1) Calculate the fitness function according to formula (3);

[0060]

[0061] In the above formula, M is the total number of iterations; i d is the stator d-axis current detection value; i q is the stator q-axis current detection value; is the estimated value of stator d-axis current; is the estimated value of the stator d-axis current.

[0062] 2) If the fitness function is 0, update the current best individual average extreme value and the current best group extreme value, otherwise, keep the original parameter value;

[0063] According to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value, the way to perturb the extreme value is:

[0064]

[0065] in:

[0066]

[0067] In the above formula, n represents the number of evolutionary stagnation times of the current best individual extreme value and the current best group extreme value, and N represents the stagnation step threshold that needs to be disturbed for the current best individual extreme value and the current best group extreme value.

[0068] The steps of obtaining the optimal estimated value of the system state at that moment through time update and measurement update include:

[0069] [1] Discrete state equations and measurement equations;

[0070]

[0071] Where: Magnetic flux and inductance identification observation vector Resistance Identification Observation Vector

[0072] [2] Expand f[x(k),u(k)] and h[x(k+1)] in a first-order Taylor series;

[0073]

[0074] in: x(k) is the state vector of the system; φ(k) is the state transfer matrix; B(k) is the control input matrix; u(k) is the input vector of the system; z(k+1) is the observation vector of the system; H(k) is the state observation matrix; w(k) is the process noise; v(k+1) is the observation noise; T S is the sampling time.

[0075] [3]Time updates include the following parts:

[0076] (a) Predict the state at the next moment based on the input and the last estimated value:

[0077]

[0078] Among them: the superscript "^" represents the state estimation value, the superscript "~" represents the predicted value; "(k|k)" represents the initial value; "(k+1|k)" represents the value of the k+1 step predicted by the k step; "(k+1|k+1)" represents the output value after the iteration ends.

[0079] (b) Prediction error covariance matrix:

[0080]

[0081] [4] Measurement update consists of the following parts:

[0082] a) Calculate the Kalman gain:

[0083]

[0084] b) Update the state variables to be identified by the system's observation vector:

[0085]

[0086] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention. All modifications and improvements based on the ideas and principles of the present invention belong to the protection scope of the present invention.

Claims

1. A method for identifying global optimal parameters of a direct-drive permanent magnet synchronous wind turbine generator, characterized in that the steps include: (1) Initialize the particle swarm, and give the initial values ​​of the position and velocity of the system noise covariance matrix Q and the measurement noise covariance matrix R; (2) Passing Continuously iterate to find the best solution to the problem, update the Q value and R value; where ν ij is the velocity vector of particle i in the jth dimension; p ij is the position vector of particle i in the jth dimension, P best-md is the current best individual average extreme value of particle i in the space search; G best-d is the current best group extreme value searched in the population; k represents the current number of iterations; c1, c2 are the state vectors of the particles; r1, r2 are random numbers between 0 and 1; w is the weight factor; (3) The current best individual average extreme value and the current best group extreme value are updated by the fitness function; (4) According to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value, the extreme value is disturbed; (5) Obtain the optimal estimate of the system state at the current moment through time update and measurement update; The step of updating the current best individual average extreme value and the current best group extreme value by the fitness function comprises: 1) According to the formula Calculate the fitness function; where M is the total number of iterations; i d is the stator d-axis current detection value; i q is the stator q-axis current detection value; is the estimated value of stator d-axis current; is the estimated value of stator d-axis current; 2) If the fitness function is 0, update the current best individual average extreme value and the current best group extreme value, otherwise, keep the original parameter value; The method of perturbing the extreme value according to the number of evolutionary stagnations of the current best individual average extreme value and the current best group extreme value is as follows: in: n represents the number of evolutionary stagnations of the current best individual extreme value and the current best group extreme value, and N represents the stagnation step threshold that the current best individual extreme value and the current best group extreme value need to be disturbed; The step of obtaining the optimal estimated value of the system state at the current moment through time updating and measurement updating comprises: [1] Establish discrete state equations and measurement equations Observation vector for flux linkage and inductance identification Resistance Identification Observation Vector [2] Expand f[x(k),u(k)] and h[x(k+1)] in a first-order Taylor series; in: x(k) is the state vector of the system; φ(k) is the state transfer matrix; B(k) is the control input matrix; u(k) is the input vector of the system; z(k+1) is the observation vector of the system; H(k) is the state observation matrix; w(k) is the process noise; v(k+1) is the observation noise; T S is the sampling time; [3] Time update includes predicting the state at the next moment and the prediction error covariance matrix based on the input and the last estimated value. The method for predicting the state at the next moment based on the input and the last estimated value is: Among them, the superscript "^" represents the state estimation value, the superscript "~" represents the predicted value; "(k|k)" represents the initial value; "(k+1|k)" represents the value of the k+1 step predicted by the kth step; "(k+1|k+1)" represents the output value after the iteration ends; The method of the prediction error covariance matrix is [4] The measurement update includes calculating the Kalman gain and updating the state variables to be identified by the system's observation vector: The method for calculating the Kalman gain is: The method of updating the state variables to be identified by the system's observation vector is:

2. The global optimal parameter identification method of a direct-drive permanent magnet synchronous wind turbine according to claim 1, characterized in that: The system noise covariance matrix Q and the measurement noise covariance matrix R can be expressed as: Where W is the process noise vector; V is the observation noise vector; E{} represents the mathematical expectation.

Citation Information

Patent Citations

  • Online identification method for permanent magnet synchronous motor

    CN109787524A

  • Parameter identification method for parallel hybrid vehicle power permanent magnet synchronous motor

    CN110266230A