Intelligent Network Model Based on Information Fusion Theory for On-line Fault Diagnosis

In order to solve effectively the issues of on-line fault diagnosis for complex equipment, an intelligent network model is newly developed for on-line fault diagnosis based on the science of cognition and information fusion theories in Internet environment. It's such a diagnostic strategy that combines not only qualitative with quantitative diagnoses but local with integral diagnoses. In this way a qualitative diagnosis shall be made first. In case the equipment is abnormal, the integrated neural network group will start immediately the quantitative analysis and classification of multi-source characteristic information from sensors, and the ES module will explain qualitatively the reasoning process of ANNs. As a result, the D-S evidence reasoning module will diagnose synthetically the output of ANNs at integrated fusion center, thus improving evidently the precision and reliability of diagnosis conclusion. The model has been tested and run on Fengman Hydropower Station's digital simulation system we developed, and the result showed that the proposed model is not only effective, versatile and applicable in practical use but beneficial to the development of local fault diagnosis system.