An Intelligent Supervisory System for the Thermal Process during Severe Disturbance

Abstract

The boiler system is a MIMO system with the characteristics of nonlinear dynamics, with there being severe loading variation, the boiler variables such as steam pressure and water level under go some major fluctuations. Such fluctuations are difficult to minimize using classical control techniques. In response to these circumstances, this paper contributes a methodology to design a two-level structure control method to improve the performance of the boiler response during such disturbance, in which the direct level consists of a predictive PID controller based on DCS(Distributed control systems); and the set-points of the drum steam pressure and the water level deviation are modified whenever necessary by a supervisor partially based on fuzzy theory according to the current operation condition. Finally, the good performance of the two-level control method is demonstrated through numerical experiments. Index Terms -Boiler system˗Supervisor controller˗Predictive PID control˗Fuzzy logic 1.INTRODUCTION The power plant is a typical process which consists of boiler, turbine and generator. But the operation mainly depends on the performance of the boiler system. The controlling to the boiler is very difficult because it is a MIMO process and has the structure instability. The classical control is to dedicate a controller for each subsystem with little attention to the coupling effects, but the performance is not satisfying. A great deal of studying has been given to provide alternatives to the classical control of power plant components. The prior study is the literature [1] which extends a general furnace follow machine method according to the compensator based on state feedback, aimed to reduce the coupling influence among the systems. Paper [2] designs a feedback controller to achieve the in-connected controlling between the steam pressure of the evaporator export and the enthalpy value.Subsequently, the self-adaptive control theory is applied in the steam temperature [3] . In this paper, the control value is the sum of the weighted variables and the weighted variables are self-adaptive control value and the classical control value, and they are exist simultaneity. This method improves the control quality and unit thermal efficiency during changing load. The researchers in our country also have gained many achievements by using self-adaptive control in this field [4] .But it is difficult to get utility controlling. In paper [5], the nerve cell and the general controller are combined. The unit load controller aimed to the linear model for the uncertain thermal power unit is designed. The control character is better then that of the classical controller. But all of these are projected based on regulatory level. It is difficult to hold up the secular benefit. From an automation point of view, Supervisory system determines process operating conditions to command the lower level automation functions. The hierarchical structure of power systems seemed to favour supervisory optimization of power units via set-point scheduli...