Predictive Functional Control And Its Application Research In Thermal Power Plant

A brief review of Model Predictive Control (MPC) is addressed firstly, which pays much attention to Predictive Functional Control (PFC), applications of MPC in thermal power plant and one of research hotpots, Distributed Model Predictive Control (DMPC). The basic principles of PFC are introduced, and PFC algorithm for first-order plus dead-time system is delivered.According to the basic principles of PFC, combining prevenient research results, novel PFC algorithms and control strategies are presented, which lead to easier design and better performances of PFC systems.A novel PFC algorithm based on Finite Response Impulse (FIR) model is presented. Analytical control laws with both one base function (step function) and two base functions (step and ramp functions) are provided. Closed-loop system steady state performance is discussed, which shows that the control system both has not remnant difference due to the variety of set point and output disturbance. The algorithm is fit for open-loop stable systems.A novel PFC algorithm for integrating plants is presented based on step response model that can be acquired easily. Analytical control law is provided. Closed-loop system steady state performance analysis show that the control system has not remnant difference due to the variety of set point, step disturbances of manipulated variable and output.For the non-minimum phase system that only has right-half plane poles, i.e. the open-loop unstable non-minimum phase system, PD feedback compensator is designed, and the PD feedback compensator and the controlled plant are treated as generalized controlled plant. PFC algorithm for first-order plus dead-time system is applied to control the generalized plant. And simulation results show its availability. Considering the non-minimum phase system that only has right-half plane zeros, i.e. the open-loop stable non-minimum phase system, a novel PFC scheme based on reference trajectory self-tuning is presented, and simulation results demonstrate its superiority.PFC algorithms are applied to thermal processes in thermal power plant, and plenty of simulation researches are carried out, which provides a solid basis for their practical applications.Combining the advantages of PFC algorithms and cascade control strategy, PFC-PID cascade superheated steam temperature systems are designed. The PFC algorithm for first-order plus dead-time system and the PFC algorithm based on FIR model are adopted respectively. Simulation results show that the superheated steam temperature system adopted PFC-PID cascade control strategy has more favorable dynamic characteristics than the system with PID cascade control strategy, and has good robustness and disturbance rejection. In order to overcome the influences of steam flow (load) on superheated steam temperature, PFC strategy with feedforward compensation for load fluctuation is provided in superheated steam temperature system, adding steam flow disturbance model to PFC prediction model. According to PFC scheme taking measurable disturbances into account, control algorithm is presented, adopting PFC algorithm for first-order plus dead-time system. Simulation results show that the system has obvious improvement in load disturbance rejection.In order to get over the influences of operating regime on superheated steam temperature dynamic characteristics, two strategies are provided: multi-model PFC strategy and fuzzy adaptive PFC strategy in superheated steam temperature system. In the first strategy, fixed models are established in some operating regimes, model prediction output is obtained with corresponding model according to system operating regime, and then manipulated input is worked out, thereby the performances of the superheated steam temperature system adopted PFC algorithm are improved once more. The basic idea of the second strategy is same as that of the first strategy. However, identified model output has strong generalization ability because of the interpolation mechanism of fuzzy model, therefore, system dynamic chara...