Adaptive PID Control Based On Model Prediction And Rolling Optimization Mechanism

PID controller plays an important role in modern industrial process control,and its control quality depends on PID parameters.The actual controlled processes may change due to the external environment,resulting in the PID control parameters are no longer suitable to the current controlled object and the control effect will become worse.So,the PID control parameters need to be tuned again.For the multivariable system,the PID parameter tuning is more difficult because of the coupling effect within the system.In addition,due to various process requirements,operators often need to switch the manual/automatic mode of control loop,so the equivalent controlled object of the PID controller will change,and the original controller parameters will no longer be suitable.Therefore,based on the mechanism of model prediction and rolling optimization,the PID parameter self-tuning problem is studied in single-variable and multi-variable systems.Finally,based on the multi-variable frequency domain design method,the PID parameter self-tuning problem in multi-loop mode switching is also studied.In order to solve the PID parameter self-tuning problem of single-variable system,the recursive identification and model transformation for controlled object are presented.Based on the identified model,the corresponding prediction model is obtained,and the closed-loop identifiability is proved when the PID controller parameters change.Based on the mechanism of model prediction and rolling optimization,the adaptive law of PID parameters is obtained by indirect or direct solving methods.For the PID parameter self-tuning problem of multi-variable system,the subsystems of multi-variable system are provided,and the online recursive identification method of each subsystem are deduced,and each subsystem model is converted into the corresponding prediction model.The closed-loop identifiability of multi-variable system is also proven when the parameters of multi-loop PID controller change.Based on the prediction model of subsystem,the indirect and direct solutions of PID parameter self-tuning for single-variable system are extended to multi-variable system,and the adaptive laws for PID parameters of each control loop are also obtained.For the PID parameter self-tuning problem of decentralized multi-loop PID controller in manual/automatic mode switching,based on the global Nyquist stability criterion,the Gershgorin circle boundary points of the first-order damp plus lag elements are introduced to quantitatively analyze the stability variation degree of each control loop before and after mode switching.The corresponding controller parameter correction scheme is designed to ensure the closed-loop stability of the whole system after mode switching.