基于子空间辨识的焦炉集气管压力控制

Coke Oven Collector Pressure Control Based on Subspace Identification

针对焦炉集气管压力具有多变量、耦合、时变性等特点,设计焦炉集气管压力增量式在线子空间多变量预测控制策略.在增量式子空间预测控制的基础上,引入滚动窗口子空间辨识方法,设计子空间预估器模型的更新策略,实现了在线子空间自适应预测控制.应用在线子空间辨识方法对焦炉集气系统现场数据进行辨识,取得了较好的预测精度;利用子空间预估器模型进一步建立焦炉集气系统的状态空间模型,在考虑输入约束、模型时变和干扰的情况下,该模型表现出了很好的控制精度和性能.

In the light of the characteristics of multi-variable, coupling and time variation of coke oven gas collector pressure, a incremental online subspace multi-variable predictive control strategy was designed for coke oven gas collector pressure. On the basis of incremental subspace predictive control, rolling window subspace identification method was introduced to design the updating strategy of subspace predictor model, and the online subspace adaptive predictive control was realized. Online subspace identification method was used to identify field data of coke oven gas collecting system and better predicted precision was obtained. State space model of gas collecting system was built based on subspace predictor model. In the condition of considering input constraints, time-varying model and disturbance, the control strategy indicated good control precision and performance for state space model of gas collector pressure.