PID Controller Parameters Optimization Based On Extremum Seeking Algorithm For Batch Process

The optimization of Proportional Integral Derivative(PID)controller parameters for batch process control has always been a hot issue in industrial applications.However,there are usually time-varying and nonlinear characteristics for batch process,which makes the optimization of PID controller parameters for batch processes difficult.Extremum Seeking Algorithm is a model-free optimization algorithm,which can obtain the optimal solution through online seeking and iterative optimization without prior knowledge of the object.In this dissertation,based on Extremum Seeking Algorithm,the parameter optimization of time-varying PID controller is researched.The main research contents of this dissertation are:(1)Based on continuous time Extremum Seeking Algorithm,the discrete form of the algorithm is obtained by using the Forward Euler Method.Aiming at the problem of PID parameters optimization,a kind of Extremum Seeking Algorithm with constraints is proposed.The algorithm considers non-negative constraints and seeking step size constraints of optimized parameters in the seeking process,which can ensure the convergence of PID parameters optimization.The effectiveness of the algorithm is verified by seeking and optimizing the PID parameters for four kinds of objects with different characteristics.(2)Considering time-varying characteristics of batch process,a kind of time-varying PID controller parameters using cubic spline interpolation is proposed,and then a kind of Extremum Seeking Algorithm with constraints is used to optimize the parameters of time-varying PID controller.The interpolation method significantly reduces the seeking dimension,which not only makes the searched parameters still retain the physical meaning,but also makes it easier to set the initial values during simulation.Non-negative constraints and seeking step size constraints of optimized parameters ensure the convergence of the seeking algorithm,and the simulation verifies that the tracking performance of time-varying PID controller using cubic spline interpolation is better than time-varying PID controller using linear interpolation and time-invariant PID controller.(3)In order to eliminate the steady-state oscillation of the system output in the steady state,a constrained Multivariable Extremum Seeking Algorithm Without Steady-State Oscillation is proposed to optimize the parameters of time-varying PID controller.Besides non-negative constraints and seeking step size constraints of optimized parameters,this algorithm also adds non-negative constraints and seeking step size constraints to the amplitude of the perturbation signal,which effectively ensures the convergence of the algorithm.By comparing the variance of the objective function and the PID controller parameters in the steady state,the effectiveness of the algorithm in eliminating the steady-state oscillation of the system output is verified.To sum up,for the PID controller parameters optimization problem,this dissertation adopts Extremum Seeking Algorithm,studies the design method of non-negative constraints and seeking step size constraints,and proposes a constrained Multivariable Extremum Seeking Algorithm Without Steady-State Oscillation,which is used for time-varying PID controller parameters seeking and optimization process for batch process.This algorithm can not only achieve the convergence of seeking optimization,but also improve the performance of tracking response,and effectively eliminate the oscillation of the system output during the steady state.