Online performance monitoring and diagnosis based on RTSID and KPLS

In the case that the control system performance is detected in poor state, it is desirable that the underlying cause can be found and the information can be used for the adjustment and recovery. Developing a recursive two-stage identification (RTSID) algorithm and a recursive control performance assessment algorithm with kernel partial least square (KPLS), then integrating the proposed identification algorithm into the performance diagnosis, this paper presents an online control performance monitoring and diagnosis strategy for the closed-loop system. The proposed algorithms are applied to the typical research example and the effectiveness of the strategy is validated.