Model-Driven PID Control System, its properties and multivariable application

PID control systems are widely used as a basic control technology in the industrial control system today. However, tuning of PID control systems is not always easy, because of its simple control structure to wide classes of industrial control processes. In order to improve the issues, we developed a ModelDriven PID control system, named MD PID Control system, by extending the “model driven control” concept proposed by Kimura(2000). Properties of the MD PID Control system are as follows; a). Two degrees of freedom property, that is, both good set-point tracking and good disturbance regulation. b). Wide applicability to various kind of industrial control processes. c). Compatibility to conventional PID Control systems by setting only control parameters. We have already successful applications in industry of Japan. Multivariable model predictive control systems are used for many industries, especially petrochemical industry, however, conventional multivariable model predictive control systems have several issues, such as slow regulating property for unexpected disturbance and hard works to maintain the multivariable process model. In this paper, we suggest a simple multivariable control system by extending Model-Driven PID control system for interacting processes and show a numerical example, after introducing the structure and properties of the MD PID control system. 1.Introduction PID control systems [1], [2] are widely used as a basic control technologies in today’s industries, as in Control Technology Survey Report 1996 in Japan [3], [4] at SICE and a Paper by Desborough and Miller [5] at Chemical Process Control 6, 2001. However, tuning of PID control systems is not always easy, because of its control limitations based on its simple control structure to wide classes of industrial control processes, such as long deadtime processes, oscillatory processes, unstable processes and processes with zero. In order to improve the issues, several advanced PID control technologies, such as, I-PD control systems by Kitamori[6,7] and Two degrees of freedom PID control systems by Araki[8], Hiroi[9] and Shigemasa[10], have been proposed since the early 1980s. However, these control systems have the same issue for long dead-time processes. Recently, ”From Model-based Control to ModelDriven Control”[11] was proposed by Kimura at the International Conference on Decision and Control (CDC) 2000 Sydney. He defined a Model-Driven Control (MDC) concept as “ a control system architecture which uses a model of the plant as a principal component of controller”. The advantages of the MDC are simple and easy to understand the control architecture, good tenability and its proved robustness. PID d τ control systems by Shinskey [12],[13], PPI (Predictive PI) control systems by Hagglund and Astrom[1] and Internal Model Control (IMC) systems by Morari and Zafiriou [14] can be considered to belong to the same class as the MDC. However, it is difficult to realize the MDC structure for unstable processes and oscillatory processes. In order to improve these issues, we proposed a ModelDriven(MD) PID control system, which is combined with PD feedback, Q filter, model and set point filter at the first ISA/ JEMIMA/SICE Joint Technical Conference 2001 Tokyo [15], the CCA/CACSD2002 Glasgow[16], the 46th JACC2003 Okayama[17]. (s) Ĝ G(s) In this paper, we suggest a simple multivariable control system by extending application of the MD PID control system for interacting processes and show a numerical example, after introducing the structure and properties of the MD-PID control system. 2. Model Driven PID Control System Fig.1 shows a block diagram of a MD PID control system, the control system consists of the following three blocks as; (a) PD feedback compensator , (b) Main controller blocks consist of a gain block, a second order Q filter with tuning parameters λ and α , and a normalized first order delay model with dead time , (c) Set-point filter , where r, v, u, y, d are set-point, internal input for local loop, an input for the process P(s), output and disturbance, respectively.