为优化1 000 MW超超临界机组协调系统控制性能，设计了基于多模型广义预测控制算法的新协调控制方法。为避免机理建模与实际情况的偏差，利用现场试验数据建立全工况的传递函数模型集，以逼近被控对象的非线性过程。新协调控制策略以多模型为基础，结合被控对象动态特性，并保留了“反馈+前馈”的控制结构。工程应用表明，多模型广义预测控制方法与传统的PID控制相比，提升了机组协调控制系统性能，同时明显改善了自动发电控制（automatic generation control，AGC）的运行考核结果。

In order to improve the control performance of the 1 000 MW Ultra supercritical unit coordinated control system,a new method based on multimodel generalized predictive control algorithm is designed.To avoid the deviation of the mechanism modeling,the transfer function model sets under all working conditions are established by using experimental data to approximate the nonlinear process of the controlled object.The new strategy is based on multimodel,which combines the dynamic characteristics of the controlled object and retains the control structure of "feedback+feedforward".Engineering application shows that compared with traditional PID control,the multimodel generalized predictive control method improves the performance of the coordinated control system and the AGC operation evaluation results significantly.

TK323

中广核研究院有限公司青锋计划项目（R-2018ZBQFC003）