Predictive Control And Its Application Research In Boiler Steam Temperature Control

The steam temperature of boiler in power plants has large inertia, large time delay and multi-factor disturbance characteristics, which makes conventional cascade PID control difficult to make steam temperature value within the scope of the ideal value, besides, it is difficult to be adaptable to Automatic Generation Control (referred to as AGC) control requirements. Even some of the advanced control strategies are powerless when the system was severely disturbed. Due to the actual project requirements of power companies, this paper have carried out the following research based on predictive control theory and its application in steam temperature control system.Firstly, steam temperature object data of different conditions and different times are collected through field tests in order to get the boiler steam temperature object model. Besides, modeling research is carried out in different needs according to several different modeling methods including traditional modeling method which is based on the step response (tangent method, the two-point method and area method) and intelligent modeling algorithm which is BP neural network modeling, RBF neural network modeling method.Secondly, the Dynamic Matrix Control (Abbreviation is DMC) and Generalized Predictive Control (Abbreviation is GPC) were deeply studied in this paper, including the simplified research of DMC and the Optimization research of GPC. An improved hybrid particle swarm optimization (IHPSO) algorithm is used to overcome the problem that the basic particle swarm optimization algorithm trapped in local optimal point easily. Simulation results show that IHPSO algorithm has higher efficiency and accuracy, better tracking performance and control quality than basic PSO algorithm, so it is suitable for online rolling optimization and solution of generalized predictive control.Third, combined with the actual project requirements, a generalized predictive control strategy which based on indirect energy balance method is presented in this paper. Through construct an indirect energy balanced correcting reduced temperature water system, changing cascade variable servo system to be constant servo system. Matlab simulation result shows that the improved method in this paper have effectively overcome the large delay characteristics of superheated steam temperature and significantly improved the control precision of boiler superheat and reheat steam temperature.Fourth, in order to apply predictive control algorithm to actual steam temperature system, the new control algorithm must be embedded into the DCS platform which is provided by the manufacturer, and generated the dynamic links and parameter configuration to update the existing modular library, and this part of work is the key to successful application of this subject. Combined with Guodian Zhishen EDPF-NT Plus DCS control system platform, the modular development research and the corresponding test of predictive control software have been done.Finally, using the successfully compiled new algorithm module to build SAMA configuration and constructing new predictive control system circuit, then offline debugging and online application on site. Lab offline simulation results show that the new proposed algorithm has better effect and good control robustness in steam temperature control. Now, the new algorithm is entering the field debugging operation stage.