本项目以具有复杂转移概率的广义Markovian跳跃系统为对象，研究鲁棒故障诊断及容错控制问题；并将上述系统推广到网络环境中，开展进一步地深入研究。不确定性和多源干扰的存在会影响故障诊断的快速性，还有可能出现误报或漏报现象，导致故障诊断单元的输出并不能正确描述系统当前运行的模态。从而增加了容错控制器设计的困难性，同时也超出了常见控制方法的研究范畴。基于此，本项目提出新颖的基于干扰观测器的鲁棒故障估计算法。在故障估计基础上，构造新颖的复合分层主动容错控制器：对故障可补偿、对干扰可补偿和抑制。.本项目部分理论成果拟于超（超）临界机组仿真平台进行验证。本项目旨在提高广义Markovian跳跃系统在实际工程中的数学建模和应用能力，丰富广义Markovian跳跃系统的故障诊断及容错控制理论体系，并为其在工程实际的应用提供新的理论支持。

The project investigates the robust fault diagnosis and fault tolerant control problem for singular Markovian jump system with complex transition probability; Based on this, we extend the above system to the network-based environment, and carry out further study. Uncertainties and multiple disturbances in the process can cause fault detection delay, missing detections and false alarms; hence the fault diagnosis output does not always indicate the true operating modes of the system. And thus, the design of fault tolerant controller becomes more difficult, and conventional control method does not work any more. According to this, a novel robust fault estimation algorithm is proposed in the project based on the disturbance observer. Based on the estimation, a novel composite hierarchical active fault tolerant controller, which can compensate fault, and reject and compensate the disturbances, is constructed..Parts of the theory of the project intend to be verified in the simulation platform of the (ultra) supercritical thermal power unit.The project aims to improve the mathematical modeling and application of singular Markovian jump system in practical engineering, rich the fault diagnosis and fault tolerant control theory of singular Markovian jump system, and provide new theoretical support to its application in the practical engineering.