由于大时滞、非线性的控制难点，典型火电机组选择催化还原（Selective Catalytic Reduction，SCR）脱硝系统的控制效果难以保障。作为一种先进控制算法，与PID控制相比模型预测控制对大时延、大惯性的被控对象有着较好的控制效果，因此在脱硝控制系统优化中多有应用。然而SCR被控对象的非线性以及未知的外部干扰仍然会影响乃至恶化模型预测控制效果，因此提出了一种改进型多模型预测控制算法。该算法具有多个模型预测控制器以适应被控对象的非线性，通过设置闭锁条件以改善模型失配时的控制效果恶化，构建了动态前馈以加速调节并克服外部干扰。最终在此基础上开发了一套火电机组SCR喷氨优化装置并在某310 MW亚临界机组上实现应用，应用结果证明了该算法的有效性和优越性。

Because of control challenges such as large time delay and nonlinearity,the performance of typical control system of SCR is hardly satisfactory.As an advanced control algorithm,model predictive control was applied for the optimization of SCR control system since its control performance for the controlled object with large time delay and inertia was better than typical control system based on PID regulator.However nonlinearity and unknown external disturbance would degrade the performance of model predictive control.Therefore an improved multiple model predictive control algorithm was developed in the paper.In the improved algorithm,multiple model predictive controllers were introduced to match the nonlinearity of controlled object,locking conditions were set for control performance improvement when model mismatch happened and dynamic feedforward loop was designed to suppress external disturbance.Finally,an optimization device based on the improved algorithm was developed and applied to an 310 MW subcritical thermal power unit.The application indicated that the algorithm was valid and superior.

X773

国家能源集团科技创新项目(GJNY-19-06);湖南省科技创新平台与人才计划(2016TP10270;湖南省湘电试验研究院有限公司科技项目(XDKY-2019-001)