摘要控制系统的复杂程度随着科技的进步与发展逐渐加大，因此，如果能够及时准确 地将系统中存在的故障诊断出来，同时采用相应的容错控制手段补偿故障产生的影 响，提高系统的可靠性和安全性变得非常重要。尤其神经网络的出现和发展，为该方 面的技术研究提供了崭新的手段和策略。83388
本文在介绍故障诊断、容错控制及神经网络的相关理论知识基础上，选择了广泛 应用的 BP 网络和概率神经网络，分别对结构复杂的非线性柴油机系统模型进行故障 诊断。并针对非线性系统提出一种故障诊断和容错控制算法，将系统故障利用概率神 经网络进行建模，达到在线估计故障向量的目的，通过引入反馈控制器和故障补偿控 制器而完成了容错控制器的设计。
由故障诊断仿真结果可得，基于概率神经网络的故障诊断在诊断速度、辨识故障 准确率等方面均优于 BP 网络，同时概率神经网络强大的非线性分类能力得到了印证； 最后利用 Lyapunov 方法对算法进行稳定性分析，算法有效性得到了证明。
毕业论文关键词：故障诊断；容错控制；神经网络；非线性系统
Abstract With the development and progress of technology, the control system is becoming more and more complex。Therefore,diagnosing the faults of the system timely and accurately,and adopting fault tolerant control technology to compensate the influence of fault to guarantee the security and reliability of the system becomes more and more important。Especially the development of artificial neural network provides a new way for the research of fault diagnosis and fault tolerant control technology。
In this paper, through the study of the relevant theoretical knowledge of fault diagnosis and fault tolerant control and neural network, the fault diagnosis of the complex nonlinear diesel engine system is carried out based on he BP network and probabilistic neural network which are widely used。Then, a fault diagnosis and fault tolerant control algorithm is put forward for the nonlinear system。The system faults are modeled using probabilistic neural network to achieve the online estimation of the fault vector。The fault tolerant controller design is completed through the introduction of feedback controller and fault compensation controller。
According to the diagnostic results,the fault diagnosis based on probabilistic neural network can be better than BP network in terms of diagnostic rate and identifying fault accuracy, and it is also proved that the probabilistic neural network has a strong ability of nonlinear classification。Finally, the stability of the algorithm is analysed using the Lyapunov method, and the effectiveness of the algorithm is proved。
Keywords:fault diagnosis；fault tolerant control；neural network；nonlinear system
目 录
第一章 绪论 1
1。1 课题研究背景和意义 1
1。2。1 故障诊断发展状况 1
1。2。2 容错控制发展状况 2
1。3 本文主要研究内容和结构安排 3
第二章 控制系统的故障诊断 5
2。1 故障诊断的定义 5
2。2 故障诊断的方法 5
2。2。1 基于解析模型的故障诊断方法 6
2。2。2 基于信号处理的故障诊断方法 基于神经网络的故障诊断与容错控制技术研究:http://www.youerw.com/zidonghua/lunwen_98305.html