PID controller tuning methods for process industry applications
A discussion of identification methods employed in process control applications is carried out. The identification methods discussed range from explicit modelling techniques based on the relay experiment and the Phase-Locked Loop methods of nonparametric system identification, through to the implicit modelling techniques of subspace identification and the model-free methods used in Iterative Feedback Tuning. For a given range of gain and phase margins, graphical methods are developed that show the viable gain margin and phase margin design pairings that are achievable by the use of a PI controller as the compensation element in a closed loop control system. Two further graphical methods that allow the parameters of a PID controller to be determined such that gain and phase margin design specifications can be met are discussed. Iterative tuning methods that allow the design of PI controllers to meet gain and phase margin specifications are developed. An extension of the iterative tuning method that allows the design of PI controllers to meet maximum sensitivity and phase margin design specifications is also discussed. The Phase-Locked Loop (PLL) method of system identification is used to carry out the closed loop identification and tuning of cascade connected control systems. The closed loop identification of multivariable systems using the PLL method of system identification and the design of a decentralised control system based on an extension to the exact gain and phase margin design method is discussed. The Iterative Feedback Tuning (IFT) method of restricted structure controller design is discussed. A new method, Controller Parameter Cycling (CPC), is introduced. The CPC method of controller tuning allows the determination of both the cost function gradient and Hessian from experiments that are carried out on the closed loop system. Thus, improved numerical techniques can be used by the CPC method over those employed in the IFT method.