论文部分内容阅读
长期以来,关于随机动态系统的故障诊断和容错控制的研究一直是控制理论和应用的重要领域之一。随机控制系统故障诊断的目标是建立有效的故障估计算法以使残差信号方差最小。这种方法仅适用于高斯型残差或者具有对称分布的概率密度函数的残差。然而,对非高斯残差而言,仅使用残差信号的方差不能够全面表示残差的不确定性。针对非高斯随机动态控制系统提出了新的故障诊断和容错控制算法,以使故障诊断中残差信号的熵极小化,同时极小化故障状态下闭环控制系统跟踪误差的熵。
For a long time, the research on fault diagnosis and fault-tolerant control of stochastic dynamic systems has been one of the important fields of control theory and application. The goal of fault diagnosis in stochastic control systems is to establish an effective fault estimation algorithm to minimize the residual signal variance. This method only applies to Gaussian residuals or to residuals with a symmetric distribution of probability density functions. However, for non-Gaussian residuals, the variance using only the residual signal can not fully represent the residual uncertainty. A new fault diagnosis and fault tolerant control algorithm for non-Gaussian stochastic control system is proposed to minimize the entropy of residual signal in fault diagnosis and to minimize the entropy of tracking error of the closed-loop control system under fault condition.