Transfer function identification method based on extended crossover genetic algorithm

Abstract

The invention provides a transfer function identification method based on an extended crossover genetic algorithm, and the method comprises the steps: obtaining a group of control system frequency domain response characteristic data through a control system frequency domain response characteristic test; a transfer function is established, and the formula of the transfer function has an undetermined coefficient: the undetermined coefficient is solved by using a genetic algorithm based on extended crossover, so that the variance of the fitted transfer function and the frequency domain response characteristic data reaches the minimum value. According to the method, the frequency domain response curve of the transfer function is gradually fitted through the extended crossover genetic algorithm, so that the purpose of identifying the transfer function is achieved, and the method can adapt to relatively complex transfer functions.

Description

Technical Field

[0001] The invention relates to the field of transfer function coefficient calculation of control systems, and in particular to a transfer function identification method based on an extended crossover genetic algorithm. Background Art

[0002] In the analysis of the characteristics of new energy control systems, the control system transfer function

[0003] The coefficients m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k It is a key parameter that directly affects the dynamic characteristics of the control system and is a key parameter of the control system. Accurately calculating the transfer function coefficient by the control system is the key to accurately calculating the control system parameters.

[0004] The currently widely used calculation method is to use the time domain response curve for identification, which cannot accurately reflect the characteristics of the control system in each frequency band. Summary of the invention

[0005] In order to accurately obtain the frequency domain response characteristic model, the present invention provides a transfer function identification method based on an extended crossover genetic algorithm, which can calculate the coefficients of the transfer function for the frequency domain response curve, realize the identification of the control system, and provide an accurate basis for the analysis and calculation of the control system.

[0006] A transfer function identification method calculation method based on an extended crossover genetic algorithm comprises the following steps:

[0007] (1) Through the control system frequency domain response characteristic test, a set of control system frequency domain response characteristic data is obtained:

[0008] X={Am i ,Ph i ,w i},i=1,2,...,m; (1)

[0009] where w i is the test frequency point, Am i is the amplitude response measured at the corresponding frequency point, Ph i is the phase response measured at the corresponding frequency point, and m is the number of points tested;

[0010] (2) Establish the following transfer function:

[0011]

[0012] where m1, m2, ..., m k ,mk+1 and n1,n2,...,n k-1 ,n k are the unknown coefficients of the transfer function;

[0013] (3) Use the genetic algorithm based on extended crossover to solve the unknown coefficients m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k , so that the variance of the fitted transfer function and the frequency domain response characteristic data in step (1) is:

[0014]

[0015] Reached minimum value.

[0016] Furthermore, the specific steps of the genetic algorithm based on extended crossover are:

[0017] 1) Generate the first generation population, the number of which is n;

[0018] 2) Crossover the n individuals to generate new m individuals, where the crossover uses the following extended crossover algorithm so that the range of the new population generated by each crossover is not less than the range of the selected population:

[0019] Z mi =Z i +3S·(Z j -Z i )-(Z j -Z i ), where S is a random number between 0 and 1, and Z i , Z j is the individual used for crossover among n individuals:

[0020] Z i =(m 1i ,m 2i ,...,m ki ,m k+1i ,n 1i ,n 2i ,...,n k-1i ,n ki ),

[0021] Z j =(m 1j ,m 2j ,...,m kj ,m k+1j ,n 1j ,n 2j ,...,n k-1j ,n kj );

[0022] 4) Merge the individuals in steps 1) and 2) into a new generation group, the number of which is m+n;

[0023] 5) Calculate the objective function value of each individual:

[0024]

[0025] 6) If the objective function does not decrease after 20 iterations, end the calculation and jump to step

[0026] 7), otherwise jump to step 2);

[0027] 7) After the calculation is completed, take the optimal individual m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k The value is the identification parameter.

[0028] Beneficial effects of the present invention:

[0029] 1. The extended crossover genetic algorithm can avoid being limited to the local optimal solution and has strong global search capabilities;

[0030] 2. Use frequency domain response data for identification, which is suitable for field test data acquisition;

[0031] 3. Based on the principle of genetic algorithm, the mean square error of the fitting curve is the smallest. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a schematic diagram of test data used in an embodiment of the present invention;

[0033] Figure 2 It is a schematic diagram of the iteration of the objective function calculated by the present invention.

[0034] Figure 3 It is a comparison diagram of the identification results of the frequency domain response and the test data of the embodiment of the present invention. DETAILED DESCRIPTION

[0035] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and specific examples of the present invention.

[0036] The embodiment of the present invention provides a transfer function identification method calculation method based on an extended crossover genetic algorithm, comprising the following steps:

[0037] (1) A frequency domain response characteristic test was conducted on a photovoltaic power generation system, and the frequency domain response characteristic curve was measured as follows: Figure 1As shown. Read the test data and get a set of frequency domain response characteristic data as follows:

[0038] Table 1

[0039]

[0040]

[0041]

[0042]

[0043] (2) Construct the following third-order transfer function:

[0044]

[0045] Among them, m1, m2,.m3 and n1, n2 are unknown coefficients.

[0046] (3) According to the extended crossover genetic algorithm, 200 first-generation individuals are randomly generated to start iterative calculation. After 356 generations of iteration, convergence is achieved and the minimum objective function value of each generation of individuals is generated as follows:

[0047] Table 2

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057] The objective function convergence diagram is as follows Figure 2 After the calculation converges, the individual with the smallest objective function is selected, and the values ​​of each item are as follows:

[0058] Table 3

[0059]

[0060]

[0061] The identified system transfer function is:

[0062]

[0063] Frequency domain response and test data comparison Figure 3 As shown in the figure, the red part is the test data, and the blue part is the frequency response of the identified system. It can be seen from the figure that the frequency domain response of the identified system is completely consistent with the test data.

[0064] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A transfer function identification method based on an extended crossover genetic algorithm, characterized in that: The steps include: (1) Through the control system frequency domain response characteristic test, a set of control system frequency domain response characteristic data is obtained: X={Am i ,Ph i ,w i },i=1,2,...,m; (1) where w i is the test frequency point, Am i is the amplitude response measured at the corresponding frequency point, Ph i is the phase response measured at the corresponding frequency point, and m is the number of points tested; (2) Establish the following transfer function: where m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k are the unknown coefficients of the transfer function; (3) Use the genetic algorithm based on extended crossover to solve the unknown coefficients m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k , so that the variance of the fitted transfer function and the frequency domain response characteristic data in step (1) is: Reached minimum value.

2. The transfer function identification method based on extended crossover genetic algorithm according to claim 1, characterized in that: The specific steps of the genetic algorithm based on extended crossover are: 1) Generate the first generation population, the number of which is n; 2) Crossover the n individuals to generate new m individuals, where the crossover uses the following extended crossover algorithm so that the range of the new population generated by each crossover is not less than the range of the selected population: Z mi =Z i +3S·(Z j -Z i )-(Z j -Z i ), where S is a random number between 0 and 1, and Z i , Z j is the individual used for crossover among n individuals: Z i =(m 1i ,m 2i ,...,m ki ,m k+1i ,n 1i ,n 2i ,...,n k-1i ,n ki ), Z j =(m 1j ,m 2j ,...,m kj ,m k+1j ,n 1j ,n 2j ,...,n k-1j ,n kj ); 4) Merge the individuals in steps 1) and 2) into a new generation group, the number of which is m+n; 5) Calculate the objective function value of each individual: 6) If the objective function does not decrease after 20 iterations, end the calculation and jump to step 7), otherwise jump to step 2); 7) After the calculation is completed, take the optimal individual m1, m2, ..., m k ,m k+1 and n1,n2,...,n k-1 ,n k The value is the identification parameter.

Images