非线性随机网络化系统缓变故障的估计与预报

With the increasing requirement of reliability and security of system and the rapid development of fault diagnosis technology, the research on fault estimation and incipient fault detection has attracted much attention. However, the incipient fault estimation and prediction problem has still lack in the thorough research. In this project, we study the problem of incipient fault estimation and prediction for nonlinear stochastic system, which is more suitable for describing general control system. Meanwhile, considering the advantages and wide application of the networked control system, the incipient fault estimation and prediction methods under the network condition are also explored. In detail, this study consists of three aspects. First, the existing nonlinear filtering algorithms are studied and improved to realize the estimation of the fault signals. Then based on the known incipient fault characteristics, the method for improving the estimation accuracy is explored. Second, on the basis of the improving estimation accuracy, the future trend of the incipient fault is further studied, so that the fault prediction can be realized. Third, the estimation and prediction methods under various networked conditions are explored, and then the influences of corresponding networked factors on the accuracy of fault estimation and prediction results are also analyzed. The theories and methods of this study aimed to explore the incipient fault estimation and prediction for the nonlinear stochastic system, and to study the influence of network factors on the results of fault estimation and prediction. This study will provide some theoretical support for the security of the actual system.

随着人们对系统可靠性和安全性要求的不断增加以及故障诊断技术的快速发展，故障估计问题以及缓变故障的检测问题已得到诸多关注。但缓变故障的估计乃至预报问题还缺乏深入的研究。本项目面向更适合描述一般控制系统的非线性随机系统，拟研究其缓变故障的估计与预报问题。同时考虑到网络化系统的诸多优点和广泛应用，进而探索网络条件下相应的估计与预报方法。首先，研究和改进已有的非线性滤波算法，以实现对故障信号的估计，并利用缓变故障特性等已知信息，探索提高缓变故障估计精度的方法。其次，在提高估计精度的基础上，进一步研究缓变故障未来的发展趋势，从而实现故障预报。最后，研究各种网络条件下缓变故障的估计与预报方法，并分析相应网络因素对其估计精度与预报结果的影响。本研究旨在探索非线性随机系统缓变故障估计与预报的新理论与新方法，进而研究网络因素对故障估计和预报的影响，为提高实际系统的安全性与可靠性提供一定的理论支持。

为了提高控制系统在网络化条件下的可靠性和安全性，以随机系统为研究对象，提出了基于事件触发的缓变故障估计及主动容错控制方法。分别针对带有不确定项的随机系统及一般的非线性随机系统，在伯努利丢包和马尔科夫丢包情况下，设计了基于事件触发的网络化缓变故障估计算法, 得到了系统状态和缓变故障的估计值，并阐明各个因素对故障估计精度的影响，给出了网络因素下故障估计误差有界性的条件，为系统在网络条件下的可靠性和安全性运行提供科学依据。同时，在缓变故障估计基础上，设计了面向随机系统的主动容错控制器，有效地消除了故障对系统的带来的影响，保证了系统的稳定运行。进一步地，研究了与故障估计密切相关的未知输入估计算法，利用输入信号的随机游走模型，针对不同的系统建立相应的增广卡尔曼滤波，从理论上严格证明了已有的未知输入算法本质上是增广卡尔曼滤波的一个极限。该结论揭示了两种不同类型滤波算法之间的内在联系，为未知输入估计问题提供新的思路，具有重要的理论和应用价值。