基于多变量广义预测控制算法的脱硝优化研究及应用

大型火力发电机组脱硝系统被控对象具有大迟延、大惯性、受干扰因素多以及不同负荷下的被控对象模型变化大等特性,传统的PID 控制不能满足控制要求。为解决这一问题,利用递推最小二乘法(recursive least square,RLS)建立了脱硝系统多变量模型,并基于多变量广义预测控制构建了脱硝系统优化控制策略,同时将该策略应用于某电厂330 MW亚临界机组。实践结果表明:无论是稳态工况还是复杂的变负荷工况,脱硝优化控制策略都能够很好地控制脱硝系统出口NOx 浓度,大大降低了NOx 浓度的波动,减少了尿素使用量,实现了脱硝系统稳定经济运行。

Research and Application of Denitrification Optimal Control Based on Multi-Variable Generalized Predictive Control Algorithms

For the denitrification control system of large thermal power unit, because of the characteristics of its controlled object, such as the large time delay and inertia, the impacts from multiple interference factors and model variations under different loads, traditional PID control strategy cannot meet the requirements. Regarding this problem, a multi-variable model of denitrification system is constructed by recursive least square method, and then an optimal control strategy for denitrification system is established based on multi-variable generalized predictive control. Meanwhile, this strategy has been applied in 330 MW subcritical unit of a power plant. From the practical results it is shown that the proposed denitration optimization control strategy based on multi-variable generalized predictive control has effectively controlled NO_x concentration at the denitration outlet and suppressed the fluctuation of NO_x concentration, whether the system is operated under steady state or complex variable load conditions. The usage of urea has also been reduced such that the stable and economic operation of the denitration system is achieved.