Research on an Improved Approach of Fault Diagnosis
Wan Jin-gan · Microelectronics & Computer · 2005
A novel approach to fault detection and diagnosis in system was presented based on information entropy and rough sets theory. It can extract rough decision rules from crude data using reduction theory. An new entropy- based cri- terion is used to measure knowledge capacity, and the change of information entropy can evaluate the status of the sys- tem; the key information in several reductions to generate the diagnosis rules was synthesized by an improved discerni- bility matrix, which can deal with incomplete an inconsistent information by utilizing redundant information sufficiently. Experimental result shows that the method not only can have a good diagnosis performance under incomplete information and solve random misinformation, but also can reduce calculation complexity and cost.