基于混合逻辑动态的过程控制实验装置故障检测

通过将故障表示为二进制变量建立混合逻辑动态故障模型，利用滚动时域估计（MHE）方法将故障检测问题转化为混杂预测控制问题。将基于MHE的故障检测方法用于三容水箱过程实验装置进行故障检测仿真研究，建立了实验装置在执行器部件故障下的混合逻辑动态（MLD）模型，研究了利用MHE对实验装置进行状态估计和故障检测的方法。对其中产生的混合整数二次规划问题则利用改进的离散微粒群（DPSO）算法进行求解。控制结果和故障检测结果表明了本文算法的有效性，更有助于对预测控制算法和故障检测技术的理解与应用。

Fault Detection of Experimental Unit for Process Contro Based on Mixed Logical Dynamica

In this paper, faults are modeled as binary variables, and a mixed logical dynamical （MLD） fault model is developed. Fault detection problem is then converted to a hybrid predictive control problem by using the moving horizon estimation（MHE） method. The MHE algorithm is applied to a three tank experimental unit. The MLD model with actuator faults is developed, and state estimation and faults detection are investigated. The mixed integer quadratic programming problem is solved by using an improved discrete particle swarm optimization （DPSO） algorithm. The control results and fault detection results show effectiveness of the MHE algorithm. It can help understand the predictive control algorithm and fault detection.