Controller performance assessment of servomechanisms for nonlinear process control systems

Abstract

Abstract—This paper aims at assessing the setpoint tracking performance of proportional-integral-derivative (PID) controllers for nonlinear single-input-single-output (SISO) process loops. A comparison is made between the actual system output and an artificial process output response derived from nonlinear system identification and a user defined closed loop transient specification. Nonlinear system identification is achieved by fitting routine operating closed loop data to nonlinear autoregressive with exogenous input (NARX) models to describe a closed loop process model and the servo model. Once the nonlinear models are established they are linearized to corresponding autoregressive with exogenous input (ARX) models where they are incorporated into a controller performance strategy. The framework will allow for control practitioners to assess the current controller setpoint tracking performance for general nonlinear systems from a transient specification point of view. Simulation studies are given to validate the efficacy of the performance assessment procedure and demonstrate that it is an effective tool when setpoint tracking is of general interest.