Data-driven Approach For On-line Auto-tuning Of Minimum Variance PID Controller

According to statistics,proportional integral differential(PID)control algorithm is the most widely used control algorithm in actual production,more than 90%of designers choose PID controller.In practice,the operator can adjust the PID controller parameters to match the target performance index,and then these adjusted parameters remain unchanged throughout the operation.However,most industrial processes have the problem of unknown model or dynamic change of process model with time.In these cases,PID parameters must be readjusted to achieve robust control performance.But the real-time manual PID setting requires high experience of control engineers and greatly increases the cost of manpower.In order to overcome this problem,the selftuning method is proposed.The automatic tuning controller can make the system respond to the request automatically to get the appropriate controller parameters.When the control indexes deviates,the operator only needs to start the self-tuning program until enough output data of the full control indexes is obtained.A self-tuning PID controller is composed of a traditional PID controller and selftuning algorithm.The performance optimization of the closed-loop control system is guaranteed by judging the output data to update the parameters of PID controller in real time.It is a challenge for designers to use self-tuning program in the design of continuous and changing control system,because the system change and self-tuning are carried out simultaneously.When the process model changes in operation or the external disturbance increases,the controller parameters should not only meet the premise of closed-loop system stability,but also complete the control index specified by the operator.A data-driven approach for on-line tuning of minimum variance(MV)PID controller is proposed in this paper for a linear system subject to stochastic disturbances,in which none of a prior knowledge or/and external excitation signals is required.The main procedure is that two different rough tuning controllers are employed and switched from one to another such that two sets of output data are collected under routine operating conditions.Subsequently,based on FCOR algorithm,the corresponding linear MV controller is estimated on-line for the linear system.The parameters of MV-PID controller is obtained so as to approximate to the estimated MV controller by means of solving an optimization problem subject to a constraint of the controller stability,where the weighted penalty function is composed of the inverse of controller parameters and the difference between the proposed controller and the minimum variance controller.By using a different selection of the weighting coefficients in the penalty function,the final tuning parameters of MV-PID controllers are determined by the practical consideration of step disturbance attenuation or sometimes the trade-off between stochastic and step signal disturbance attenuation.