基于多变量广义预测控制算法的两级过热汽温协调联动控制

大型火力发电机组过热汽温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，同时面临着煤质多变、环境多变等各种复杂恶劣的工况，以传统PID控制和单变量广义预测控制为基础的汽温控制效果有限。利用递推最小二乘法（recursive least square，RLS）建立了过热汽温系统3个典型工况下的模型，并基于多变量广义预测控制构建了两级过热汽温联动控制优化策略，同时将该策略应用于某电厂330 MW亚临界机组。结果表明，无论是稳态工况还是复杂的变负荷工况，优化控制策略都能够很好地控制过热器出口汽温。

Coordinative Two-Stage Superheated Steam Temperature Linkage Control Based on Multivariable Generalized Predictive Control Algorithm

In addition to their own characteristics such as long duration of time delay and large inertia, the controlled objects of superheated steam temperature in large thermal power generating units are affected by a considerable number of factors and model variations under different loading conditions. Especially with complicate coal quality or under volatile environment, the effect of superheated steam temperature control turns out to be very limited, which are based on traditional PID control and single variable generalized predictive control. In this paper, the models of superheated steam temperature system under three typical working conditions are established by virtue of recursive least square method, and then an optimization strategy of two-stage superheated steam temperature linkage control based on multivariable generalized predictive control (MGPC) is constructed. Meanwhile, the strategy is applied to the 330 MW subcritical units in a power plant. The practice results show that the optimal control strategy is effective to control the outlet steam temperature of superheater whenever the system is operated under stable loads or complicated variable loads.