论文部分内容阅读
针对控制系统性能评估中的模型失配问题,提出了一种新的闭环模型评估方法。该方法首先分析了控制系统中扰动是具有反馈不变性的,其次推导了可以用常规闭环回路数据估计扰动序列,并利用扰动序列的反馈不变性原则定义了模型质量指标,建立该指标的时间序列模型对其进行预测,对预测残差进行统计过程控制,实现了对控制系统性能的实时评价。当检测出控制系统性能出现下降时,通过基于距离的判别分析方法快速定位性能下降源。最后通过Wood-Berry二元精馏塔上的仿真研究验证了所提方法的有效性。
Aiming at the problem of model mismatch in the performance evaluation of control system, a new closed-loop model evaluation method is proposed. The method first analyzes the perturbation of the control system with feedback invariance, and then deduces that the perturbation sequence can be estimated by using the conventional closed-loop loop data. The model quality index is defined by the principle of feedback invariance of the perturbation sequence, and the time series of the indicator is established The model predicts it, carries on the statistical process control to the forecast residual, realizes the real-time appraisal to the control system performance. When the performance of the control system is detected to drop, the source of performance degradation can be quickly located by distance-based discriminant analysis. Finally, the simulation results on the Wood-Berry binary distillation column verify the effectiveness of the proposed method.