论文部分内容阅读
基于模型的故障诊断往往采用人工智能技术来处理不确定的知识和不完整的信息。概率推理法是一种处理不确定或不完整信息的方法,而贝叶斯网络是一种能够将它应用于实际的工具。提出了一种基于故障树和键合图理论来构建贝叶斯网络模型的新方法,并对系统进行故障诊断。实现了对引起系统或过程的异常行为的元件进行准确定位,并获得同时出现故障的元件对系统影响程度的大小,即为系统操作员提供了一个关于系统组件的优先级检查和维护计划。最后,对此方法的性能进行了仿真验证。
Model-based fault diagnosis often uses artificial intelligence techniques to deal with uncertain knowledge and incomplete information. Probabilistic reasoning is a way to deal with uncertain or incomplete information, and Bayesian network is a tool that can be applied to it. A new method for constructing Bayesian network model based on fault tree and bond graph theory is proposed and the system is diagnosed. The accurate positioning of the components causing the abnormal behavior of the system or the process and the influence of the components that fail at the same time on the system are realized. That is, the system operator is provided with a priority inspection and maintenance plan for the system components. Finally, the performance of this method is verified by simulation.