基于角度修正迭代学习的离散时变系统故障诊断

Novel fault diagnosis for discrete time-varying system based on angle correction iterative learning

针对一类非线性离散时变系统的故障诊断问题,提出了一种新的故障检测与估计算法.该算法在选取的优化时域内,利用残差信号通过迭代学习方法调整引入的虚拟故障,并利用实际输出和故障跟踪估计器输出向量空间的角度关系,来修正虚拟故障的迭代学习律,以此来加快算法的收敛速度.该算法不仅能够实现不同类型故障信号的检测与估计,而且还充分利用了估计器输出信号中的新信息,使得算法的收敛速度得到明显提高.最后仿真结果验证了该方法的有效性.

A novel fault detection and estimation algorithm is proposed to solve the fault diagnosis problem for a class of nonlinear discrete time-varying system. By using the residual signal, the algorithm adjusts the introduced virtual fault through an iterative learning procedure in the selected optimization time-domain. To speed up the algorithm convergence, the angle relationship between the actual output and the output in the vector space of the fault tracking estimator is employed to modify the iterative learning law for the virtual fault. The proposed algorithm not only detects and estimates different- types of faults but also makes full use of the update information of the estimator output signal to effectively improve the algorithm convergent rate of the algorithm. Simulation results verify the validity of the algorithm.