Autotune identification under load disturbance

Frequent and large load changes are often encountered in chemical processes. This can be lead to erroneous results in system identification especially for the fully automated identification procedure in autotuners. Relatively speaking, the relay feedback test-is less sensitive to load disturbance (e.g., as compared to the step test). However, as the magnitude of load increases, the estimates of ultimate gain and ultimate frequency deteriorate exponentially. In this work, the causes of errors are analyzed using a describing function and a remedial action is devised. Since load disturbance leads to asymmetric output responses, an output-biased relay is employed to restore symmetric output response. Furthermore, instead of a trial-and-error procedure, the bias value can be calculated immediately following system responses. Subsequently, an identification procedure is proposed to maintain the quality of the model in the face of load changes. Linear and nonlinear systems are used to test the effectiveness of the proposed procedure. The results show that the proposed approach maintains the quality of estimates under step-like and non-step-like load changes.