Research On The Unit Plant CCS Based On Multiple Model Adaptive Control

Large unit power coordinated control system is important part to improve economic power and achieve power grid dispatching automation. The coordinated control system is a complex multi-variable control system, there is uncertain interference and nonlinear characteristics in the system, with strong coupling characteristics, and there is a considerable delay in the boiler side of the net , so the design of controller is difficult.Robust control is common method to deal with uncertainty. Robust control is very effective to small range of uncertainty. It can not guarantee quality of control system when parameters of the system change or large change; traditional adaptive controllers are often designed based on a fixed or slowly time-varying system model. The controller is poor for a complex controlled system.In this context, it is proposed by multiple model adaptive control. Use multiple models to approximate dynamic characteristics of the system, and design the appropriate sub-controllers for each sub-model. And by switching optimization performance or weighted combination the controller designed based on the best model is mapped to the current controller. The controller can achieve better control accuracy, tracking speed and stability for complex systems.For parameters' uncertainties, nonlinear, large delay and large inertia in the power plant coordinated control system, the algorithm of multiple model adaptive control is applied on coordination control system. First, process the object of coordination control system. Under premise of reasonable simplification, it is generally simplified as a two-input and two-output controlled object for the drum boiler power unit. The algorithm of feedforward compensation decoupling can be used to lift coupling of control loop or system variables. Optimal model reduction algorithm is used for high-level object model. Second, multiple model adaptive control is designed. The equivalent model in two operating conditions are chosed as fixed models, and then an adaptive model is chosed to ensure identification error to zero, so that the entire sub-model can be covered as dynamic characteristics of the object. Self-tuning PID are designed for models as sub-controllers. Based on the algorithm of weighted performance index, it makes a weighted of all sub-controller's output as actual value of the controller's output to control the actual object. Finally, the proposed algorithm is applied to coordinated control system. The simulation results show effectiveness of multiple model adaptive control, to pave the way for further study of multiple model adaptive control in nonlinear systems, and point out future research directions.