CN-122001253-A - Motor active disturbance rejection control method based on improved particle swarm-bacterial foraging

CN122001253ACN 122001253 ACN122001253 ACN 122001253ACN-122001253-A

Abstract

An electric motor active disturbance rejection control method based on improved particle swarm-bacterial foraging belongs to the field of electric motor control. The method comprises the steps of determining an absolute value integral ITAE function as an optimization target, screening parameters such as ESO gain, NLSEF gain, nonlinear factors, control coefficients, linear interval width and the like in a motor rotating speed-current loop second-order ADRC control link to form a vector X to be optimized, adopting an improved PSO-BFO algorithm to realize parameter optimization, updating particle speed and position through PSO, introducing PSO average speed to guide BFO to execute foraging, combining an individual optimal pbest and a global optimal gbest updating mechanism, outputting an ADRC optimal parameter by taking the J value change rate of continuous twice gbest as a termination condition that the J value change rate is smaller than a set error epsilon or the maximum iteration number T is reached, and realizing motor self-interference control according to the ADRC optimal parameter, and improving motor rotating speed tracking precision, dynamic response speed and interference rejection capability.

Inventors

DONG XUEMING
WU YINFENG
WANG YUANJIA
WANG XIAOYUN

Assignees

中国航空工业集团公司北京长城计量测试技术研究所

Dates

Publication Date: 20260508
Application Date: 20251202

Claims (3)

1. The motor active disturbance rejection control method based on improved particle swarm-bacterial foraging is characterized by comprising the following steps of: the method comprises the following steps of taking the motor rotating speed as a control object, determining an absolute value integral ITAE function as an optimization target, wherein the expression is as follows: wherein t is time, omega * (t) is reference input of rotation speed control, and omega (t) is actual rotation speed output of the motor; Step two, establishing a rotating speed-current loop second-order active disturbance rejection control link of the motor, taking omega * (t) as input, taking a motor q-axis control current iq * as output, taking the actual rotating speed omega (t) of the motor and an inverter output current iq as feedback, inputting omega * (t) into a tracking differentiator TD shown in a formula (2), and generating a smooth tracking signal x 1d of omega * (t) and a differential signal x 2d of x 1d : Wherein h is an integral step length, fst (fourth) is a selected nonlinear maximum speed control integrated function, iq is input to an extended state observer ESO shown in formula (3) for estimating an actual rotation speed and a total disturbance received by the actual rotation speed, and the expression is as follows: Wherein, the As a parameter of the state space, An estimated value of the actual rotation speed omega (t), An estimated value of the derivative of ω (t), For the estimation of the unknown disturbance to which the actual rotational speed is subjected, Representing the estimated rotation speed error, beta 1 、β 2 、β 3 is the gain parameter of rotation speed error, alpha 1 、α 2 、α 3 is the nonlinear factor, delta is the linear interval width, b 0 is the control coefficient, fal is the nonlinear function, and the estimated error of omega * (t) is calculated respectively Error rate of change of ω * (t) E 1 、e 2 is input to a nonlinear state error feedback control law NLSEF shown in equation (4): iq * ＝β 01 fal(e 1 ,α 01 ,δ)+β 02 fal(e 2 ,α 02 ,δ) (4) Wherein, beta 01 、β 02 is NLSEF gain parameter, alpha 01 、α 02 is NLSEF nonlinear factor; step three, determining control parameters to be optimized, namely ESO gain parameters beta 1 、β 2 、β 3 , NLSEF gain parameters beta 01 、β 02 , nonlinear factors alpha 01 、α 02 , control coefficients b 0 and linear interval width delta as parameters to be set, wherein 9 parameters are marked as vectors X= [ beta 1 ,β 2 ,β 3 ,β 01 ,β 02 ,α 01 ,α 02 ,δ,b 0 ]; And step four, according to the vector X determined in the step three, updating the positions and the speeds of the particle swarms according to a formula (5) by combining 4.2, 4.3 and 4.4 to perform updating operation on the pbest, the X b and the gbest and combining the updated pbest, the updated X b and the updated pbest gbest: step 4.1 for the t-th iteration, updating the position and the velocity of the particle swarm according to formula (5): Wherein c 1 、c 2 is a learning factor, r 1 、r 2 is a random number in the [0,1] interval at each iteration, pbest (j, t) is an individual optimal position in the t-th iteration, gbest (j, t) is a population global optimal position in the t-th iteration, w (t) is an inertial weight, X p (i, j, t) is a particle swarm position in the t-th iteration, X b (i, j, t) is a bacterial swarm position in the t-th iteration, V p (i, j, t) is a particle swarm velocity in the t-th iteration, D represents a population size, the same length as vector X, and N p is a population individual number; Step 4.2 updating pbest, namely sequentially extracting column vectors from X p (i, J, t+1) according to the ascending order of the parameter i and assigning the column vectors to the vectors X, controlling the operation of a motor as control parameters, recording data omega * (t) and omega (t), respectively calculating J values according to a formula (1), and assigning the vectors X with the smallest corresponding J values to pbest (J, t+1); Step 4.3 update X b for the t-th iteration, bacterial foraging is performed according to equation (6): Wherein step is the chemotactic step length, and the average speed of the particle swarm on the j-th dimensional parameter in the t-th iteration " The integral movement trend of the particle swarm in the dimension is reflected, delta (j) is the j-th component of the random direction vector, and lambda is a random number on the interval of [0,1 ]; Step 4.4 updating gbest, if the J value of pbest (J, t+1) is smaller than gbest (J, t), then gbest (J, t+1) is updated to pbest (J, t+1), otherwise, it is updated to gbest (J, t); And fifthly, performing algorithm iteration with the J value minimized as a target until the iteration is finished, wherein the global optimal position gbest is the optimal control parameter, and realizing motor active disturbance rejection control based on improved particle swarm-bacterial foraging according to the optimal control parameter.
2. The method of claim 1, wherein the fifth implementation method is, Step 5.1, initializing an algorithm, namely generating a random initial population, wherein the value ranges of a particle swarm initial position X p (i, j, 0) and a bacterial population initial position X b (i, j, 0) are consistent, setting the initial particle swarm speed V p (i, j, 0) to be 0, setting the maximum iteration number T, the population scale N p , the population dimension D, the chemotactic step size and the inertia weight w (T), and initializing optimal positions pbest (j, 0) and gbest (j, 0) to be X p (0, j, 0); step 5.2, algorithm iteration, namely executing algorithm iteration according to the iteration method of the step four; And 5.3, ending the iteration, namely when the J value change rate corresponding to gbest of two continuous iterations is smaller than the set error epsilon (J t+1 -J t )/J t < epsilon, or when the iteration reaches the maximum number T, ending the iteration, wherein the global optimal position gbest is the optimal control parameter, and realizing the motor active disturbance rejection control based on improved particle swarm-bacterial foraging according to the optimal control parameter.
3. The method of claim 1 or 2, wherein the inertial weight w (t) is linearly decreasing in the form of w(t)=w max -(w max -w min )×t/T Where w max is the maximum weight, w min is the minimum weight, w max >w min >0, and t is the current iteration number.

Description

Motor active disturbance rejection control method based on improved particle swarm-bacterial foraging Technical Field The invention relates to a motor active disturbance rejection control parameter setting method based on an improved particle swarm-bacterial foraging algorithm, and belongs to the technical field of motor control. Background In the motor control system, the accuracy, dynamic response speed and disturbance rejection capability of the rotation speed control directly determine the motor running performance. The Active Disturbance Rejection Control (ADRC) technology is one of the core schemes of motor rotation speed control because of the advantages of being independent of an accurate model of a controlled object and being capable of actively estimating and compensating internal and external disturbance of a system. However, the control performance of the ADRC is highly dependent on the setting effect of the module parameters, and if the parameter setting is unreasonable, the requirement of high-precision control of the motor cannot be met. Although the single intelligent algorithm (such as particle swarm optimization PSO and bacterial foraging optimization BFO) can realize automatic optimization of parameters, each intelligent algorithm has short plates, for example, the PSO algorithm is easy to fall into local optimization due to improper inertia weight setting, so that an optimizing result cannot cover a global optimal region of the parameters, the BFO algorithm has stronger local searching capability, but the global convergence speed is low, and the problems of excessive iteration times and low engineering application efficiency are easy to occur in high-dimensional optimizing scenes such as ADRC parameters (usually containing 9 parameters to be set) of a motor. Disclosure of Invention The invention aims to solve the problems that the existing parameter setting precision is low and the control requirement on a real motor cannot be met, and provides a motor active disturbance rejection control method based on improved particle swarm-bacterial foraging, which can give consideration to ADRC parameter setting of global optimizing capability and convergence speed, and by fusing the advantages of different intelligent algorithms, the high-efficiency and accurate setting of motor ADRC key parameters is realized, the high-precision rotating speed control requirement of a motor under complex working conditions is met, and the rotating speed tracking precision, dynamic response speed and disturbance rejection capability of the motor are improved. The invention aims at realizing the following technical scheme: The invention discloses a motor active disturbance rejection control method based on improved particle swarm-bacterial foraging, which comprises the following steps: the method comprises the following steps of taking the motor rotating speed as a control object, determining an absolute value integral ITAE function as an optimization target, wherein the expression is as follows: wherein t is time, omega * (t) is reference input of rotation speed control, and omega (t) is actual rotation speed output of the motor; And step two, establishing a second-order active disturbance rejection control link of a rotating speed-current loop of the motor, taking omega * (t) as input, taking a q-axis control current iq * of the motor as output, and taking an actual rotating speed omega (t) of the motor and an output current iq of an inverter as feedback. Omega * (t) is input to the tracking differentiator TD shown in equation (2), generating a smoothed tracking signal x 1d of omega * (t) and a differentiated signal x 2d of x 1d: Where h is the integration step size and fst (-) is the selected nonlinear fastest control synthesis function. The iq is input to an extended state observer ESO shown in the formula (3) for estimating the actual rotation speed and the total disturbance to which it is subjected, the expression being: Wherein, the As a parameter of the state space,An estimated value of the actual rotation speed omega (t),An estimated value of the derivative of ω (t),For the estimation of the unknown disturbance to which the actual rotational speed is subjected,And beta 1、β2、β3 is a rotation speed error gain parameter, alpha 1、α2、α3 is a nonlinear factor, delta is a linear interval width, b 0 is a control coefficient, and fal is a nonlinear function. Calculating the estimation errors of omega * (t) respectivelyError rate of change of ω * (t)E 1、e2 is input to a nonlinear state error feedback control law NLSEF shown in equation (4): iq*＝β01fal(e1,α01,δ)+β02fal(e2,α02,δ) (4) Wherein, beta 01、β02 is NLSEF gain parameter, alpha 01、α02 is NLSEF nonlinear factor; step three, determining control parameters to be optimized, namely ESO gain parameters beta 1、β2、β3, NLSEF gain parameters beta 01、β02, nonlinear factors alpha 01、α02, control coefficients b 0 and linear interval width delta as parameters to be set, whe