事件驱动异构多传感器网络化系统故障诊断及可靠性研究

Fault diagnosis and prognosis is an effective way to improve the system security and reliability, reduce the risk of accidents, and achieve the condition based maintenance. Existing researches have been focused on traditional time-driven systems. With the widespread use of sensor network, the contradiction between the rapid growth of data communication and the limited network bandwidth is becoming more and more prominent. At the same time, a large number of data transmission is the main factor of energy consumption of sensors. Even-triggered data scheduling can effectively reduce the amount of data transmission, and prolong the system life under the premise of desired system performance. However, the introduction of event driven mechanism brings new problems and challenges to traditional fault diagnosis and prediction. To solve above problem, this project is intended to study the problem of multi-source heterogeneous information fusion for event-triggered networked systems, and the fault diagnosis and reliability prediction problems of event-triggered networked systems based on multi-source heterogeneous information fusion. Meanwhile, simulation and experimental verifications of proposed theories and methods will be given. The research results of this project will provide new methods and theoretical basis for fault fusion diagnosis and prognosis of event-triggered networked systems, and could be widely applied to fault diagnosis and reliability prediction problems of practical networked systems, especially heterogeneous multi-sensor networked system with communicational resources and energy constraints.

故障诊断和预测是提高系统安全性和可靠性，降低事故风险，实现视情维护的有效途径。已有研究主要针对传统的时间驱动系统。随着传感器网络的广泛使用，网络中急剧增长的数据通信量与有限的网络通信带宽之间的矛盾日益凸显。同时，大量的数据发送是消耗传感器能量的主要因素。基于事件驱动的数据调度可以在保证系统性能的前提下减少数据发送量，从而有效降低通信量，延长系统寿命。然而，事件驱动机制的引入给传统的故障诊断和预测带来了新的问题和挑战。本项目拟针对上述问题，研究事件驱动下网络化系统的多源异构信息融合问题，以及事件驱动下基于异构多源信息融合的网络化系统故障诊断和可靠性预测问题，并对所得到的理论和方法进行仿真和实验验证。本项目的研究成果将为事件驱动下的网络化系统故障融合诊断和预测提供新的方法和理论依据，并可广泛应用于实际网络化系统，特别是通信资源和能量受限的异构多传感器网络化系统的故障诊断和可靠性预测。

故障诊断及预测是提高系统可靠性、保障系统安全运行的有效途径。随着传感器技术的发展，用于复杂系统状态监控的传感器的种类和数量日益增多。同时，事件驱动机制能够有效减少网络化系统中的数据传输，从而节省有限的通信资源和传感器能量。本项目围绕事件驱动机制下异构多传感器网络化系统的信息融合、故障诊断及可靠性问题开展了系统深入系统的研究。构建了异构多传感器网络化系统融合估计的问题描述框架，提出了确定性事件驱动和随机事件驱动下的分布式信息融合理论和方法。进一步，将上述结果推广到了随机混杂系统，并提出了异构多传感器网络化系统的一致性融合估计理论和方法。针对异构多传感器复杂网络化系统的乘性执行器、传感器和过程故障，提出了基于异构多传感器多模型状态融合估计的故障联合检测、分离和估计方法，能够有效处理事件驱动所引起的非线性，并有效融合多个异构传感器所提供的信息。针对异构多传感器网络化系统中存在的性能随机退化现象，提出了基于多源异构信息融合的系统剩余使用寿命及可靠性预测方法，所提出的方法能够有效利用多源异构传感器的实时测量信息，实现对随机性能退化的在线建模和系统剩余使用寿命的预测。项目的研究成果提供了异构多传感器网络化系统的信息融合、故障诊断和可靠性分析理论和方法，并可广泛应用于各类复杂网络化系统，特别是通信资源和能量受限的异构多传感器网络化系统的状态估计、故障诊断和可靠性预测。