基于协方差指标预测的MPC实时性能监控

Real-time Performance Monitoring of MPC Based on Covariance Index Prediction

为利用过程数据实时监控模型预测控制(Model predictivecontrol,MPC)的性能,提出一种基于协方差预测残差的性能监控方法.首先在分析模型预测控制器优化函数和控制结构的基础上,构造包含预测误差、控制量和过程输出的监控变量集,然后利用滑动时间窗口建立基于协方差的实时性能评价指标.针对协方差指标缺少控制限的问题,建立实时协方差指标的时间序列模型,根据协方差指标的预测残差检测模型预测控制性能下降.进一步利用基于数据集相似度的性能诊断方法确定性能恶化源.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 monitoring model predictive control performance real-timely by using process data,a covariance prediction error based performance monitoring method is proposed.On the basis of analyzing MPC optimal objective and control structure,a monitored variable set composed of prediction errors,manipulated variables and process output variables is developed.Then,a covariance based real-time performance assessment index is presented by adopting a moving window.For the problem that covariance index has no control limits,a time sequence model for real-time covariance index is presented.The predictive residual of covariance index is monitored to detect MPC performance deterioration.The source of performance deterioration can be located by using a performance diagnosis method based on data set similarity.Simulations on the WoodBerry binary distillation column demonstrate the effectiveness of the foregoing scheme.