信息融合技术在多Agent故障诊断中的应用

The Application of Information Fusion Technology to the Multi-Agent Fault Diagnosis System

由于分布式多Agent故障诊断系统中各Agent知识和经验的限制,提供的结果和可信度将有所不同而很难得到一个正确解。所以在信息融合技术的基础上,提出了解决这类问题的四种信息融合方法。模糊信息融合方法、贝叶斯信息型融合方法、D-S证据理论信息融合方法和可信度信息融合方法。这些信息融合方法功能互补、相互配合协助多Agent故障诊断专家系统将来自不同Agent的、带有差异的结果进行融合,使融合后的结果比单个Agent的诊断结果可信度更高,从而提高故障诊断专家系统诊断结果正确率。

For the limits of the knowledge and experience of the Agent in the Fault Diagnosis System of Distributed Multi- Agent. The results and credibility which it provided will be so different that it＇s very difficult to get a correct solution. Therefore based on the technology of information fusion, we proposed four methods of information fusion to solute this problem. Fuzzy Information Fusion Method, Bayesian information-based fusion method, D-S evidence theory of information fusion method and The credibility of information fusion method. These information fusion methods can help with each other to assist the Expert System of multi-Agent Fault Diagnosis to fuse the different results which come from the different Agent, so that the new result will has higher credibility than a single Agent＇s result, so as to enhance the precision of the Fault Diagnosis Expert System.