Adaptive PID Controller Design for Nonlinear Systems

In this paper, an adaptive proportional-integral-derivative (PID) controller design is proposed for nonlinear process control. In the proposed method, a controller database is constructed to store the PID parameters previously implemented together with their corresponding information vectors, while another modeling database is utilized by the just-in-time learning technique for modeling purpose. The PID parameters are obtained from controller database according to the current process dynamics characterized by the information vector at each sampling instance. Moreover, these PID parameters can be further fine-tuned, whenever required, and the resulting PID parameters together with the associated information vector are stored into controller database to improve it for operating conditions where the relevant information is not available in the construction of initial controller database. Simulation results are presented to illustrate the proposed adaptive PID design and its effectiveness is evaluated through comparison with its conventional counterpart.