许多实际系统要求控制输出概率密度函数的形状，称此类系统为随机分布系 统。随机分布系统即能表述高斯系统，又能表述非高斯系统。本项目对非高斯奇异随机分布 系统的主动容错控制进行深入的研究。研究内容主要着重如下四个方面： 1．考虑外扰和模型不确定性的影响，将分别提出基于滑模观测器和自适应迭代学习观测器的故障诊断算法，基于故障诊断的信息进行控制器重组，并分析容错控制后系统的稳定性。 2．构建基于统计信息并结合故障诊断信息的容错跟踪控制框架。 3．在研究内容1、2基础上，考虑时滞的影响，对非高斯时滞奇异随机分布系统的主动容错控制进行深入的研究。 4．当目标概率密度函数未知时，对非高斯奇异随机分布系统拟进行最小熵容错控制设计。 在理论方面，拓宽了随机分布系统故障诊断与容错控制的研究领域。所提出的算法将在东北 大学流程工业综合自动化国家重点实验室的磨矿过程半实物仿真平台上进行可行性应用验证。

In many practical control systems, the shape of the probability density function (PDF)of process variables needs to be controlled. Such systems can be described by the stochastic distribution control (SDC) system, which can not only represent Gaussian systems, but also describe non-Gaussian systems. In this project, active fault tolerant control for non-Gaussian singular stochastic distribution sytems will be researched deeply. The research contents mainly focus on the following four aspects: 1. Considering the effect of the disturbance and model uncertainties, sliding mode observer and adaptive iterative learning observer based fault diagnosis algorithms will be proposed for non-Gaussian nonlinear singular stochastic distribution systems,respectively. Based on the fault estimation information, the controller will be reconfigured to make the post-fault PDF still track the given PDF. The stability analysis after the fault tolerant control should also be carried out. 2. A framework of fault tolerant tracking control based on statistical information and the fault diagnosis information will be constructed for the non-Gaussian singular stochastic distribution systems. 3. Based on research contents 1 and 2, considering the effect of time delay, active fault tolerant control for the non-Gaussian time-delayed singular stochastic distribution systems will be researched deeply. 4. When the objective PDF information of the non-Gaussian singular stochastic distribution system is not known, the fault tolerant controller will be designed using the minimum entropy principle. Theoretically, the research area of fault diagnosis and fault tolerant control for stochastic distribution systems will be broadened. The algorithms of this project will be used in the semi-physical simulation platform of the grinding processes to test the application feasibility in State Key Laboratory of Synthetical Automation for Process Industries of Northeastern University.