The expert syste ms are constructed based on the way of human e xperts solvingproble ms. It is very mea ningful to build a hardware and software fa ult dia gnosisexpert system, because it can he lp the commo n comp uter users to d iagnostichardware and software fa ilures.But trad itiona l expert syste ms are ver y diffic ult to obtain comp lete knowledge,and the expert syste m could give the correct decis io ns only whe n it has completeknowledge. Rough set method can obtain knowledge fro m large a mounts of datawithout any prior i information, so it can solve the proble m that expert syste ms is hardto obta in comp lete knowledge. The ma in works are as fo llows:(1) A model o f the hardware and software fault d iagnosis expert syste m wasproposed, whic h constituted by knowledge base, inference engine, knowledgeacquis ition, comprehends ive database and man-machine inter face.(2) Production rules were used to represent fault d iagnosis r ules in the knowledgebase. The knowledge base stores fault dia gnos is rules and solutions for fault.(3) The reasoning strategy o f the inference engine is a forward and reverse mixedreasoning strategy. In this reasoning strategy, the forward reasoning process is to findthe possib le causes as predictions o f diagnostic results, and the reverse reasoningprocess is to ver ify the correctness of the predictions.(4) The knowledge acquis ition module was the focus of this expert system, and arough set-based knowledge acquis ition method was used in the module. Thisknowledge acquis itio n method comb ines discernib ility matrix-based attributereduction algor ithm and heur istic va lue reductio n algor ithm, which obtainsknowledge by reduct ing the decis io n tab le build by fault samp les. The exa mp le ofknowledge acquis itio n proved the feasib ility of the knowledge acquis itio n method,and the experiment proved that the reduction a lgor ithm has good time effic iency.(5) The hardware and software fault d iagnos is expert system was imp le mented bycoding. Syste m testing s hows that the syste m has good va lid ity and reliab ility. |