火电机组多变量预测控制系统的开发与应用

Study and Application of Multivariable Predictive Control System for Power Unit

针对目前我国火电机组对于先进控制软件的迫切需求,结合子空间模型辨识和预测控制技术开发出一种基于状态空间模型的火电机组多变量预测控制系统;该系统采用子空间辨识方法离线辨识状态空间模型,利用多模型切换适应火电机组部分控制回路的非线性特征,最后使用预测控制器完成对象的在线控制;在电厂协调控制回路的实际应用中,负荷升降速率4%,压力偏差小于0.18MPa。所以该系统能够有效提高控制回路的响应速度和稳定性,具有一定的实用性和推广价值。

Aiming at the urgent demand for advanced control software in the power plants, this paper developed a new type of multivari- able predictive control system based on state--space model with subspace identification and predictive control technology. The system identi- fied state--space model with subspace identification off--line, conquered the non--line character of object with multi model switching, and controlled device on--line with predictive controller. In the application of power plant, the respond speed reached 4 % and pressure deviation less than 0. 18MPa. So this system can improve stability of control loop effectively. The technology and methods adopted in the system are practical and worthy of using abroad.