Title: Performance assessment and tuning of PID control using TLBO: the single-loop case and PI/P cascade case

Abstract: Proportional-integral-derivative (PID) control, the most common control strategy in the industry, always suffers from health problems resulting from external disturbances, improper tuning, etc. Therefore, there have been many studies on control performance assessment (CPA) and optimal tuning. Minimum output variance (MOV) is used as a benchmark for CPA of PID, but it is difficult to be found due to the associated non-convex optimization problem. For the optimal tuning, many different objective functions have been proposed, but few consider the stochastic disturbance rejection. In this paper, a multi-objective function simultaneously considering integral of absolute error (IAE) and MOV is proposed to optimize PID for better disturbance rejection. The non-convex problem and multi-objective problem are solved by teaching-learning-based optimization (TLBO). This stochastic optimization algorithm can guarantee a tighter lower bound for MOV due to the excellent capability of local optima avoidance and needs less calculation time due to the low complexity. Furthermore, CPA and the tuning method are extended to the PI/P cascade case. The results of several numerical examples of CPA problems show that TLBO can generate better MOV than existing methods within one second on most examples. The simulation results of the tuning method applied to two temperature control systems reveal that the weight of the multi-objective function can compromise other performance criteria such as overshoot and settling time to improve the disturbance rejection. It also indicates that the tuning method can be utilized to multi-stage PID control strategy to resolve the contradiction between disturbance rejection and other performance criteria.