Development of an equipment failure identification expert system with multiple reasoning approaches

The goal of the reasoning system in this study is to identify the most similar failure type or failure cases. As a user inputs all possible requirements (attributes), the inference engine of the system carries out its similarity assessment (inference approaches) and ranks rules or cases from the data base. Various inference approaches are chosen to find out the optimal method for the RBR and CBR system. The CBR system offers two types of inference methods which are hierarchical factors, flat factors without weight. For RBR system, there three types of inference methods are chosen, one is complete matched and the others are partial matched approaches which use the inference capability of CBR. The performance of developed system is then evaluated by using the real cases from China Steel Corporation (CSC). For the RBR system, performance is directly check the inferred order of the document ranking list. For the CBR system, the effectiveness of each inference method is evaluated by using "Recall", "Precision", and "F-Measure" approaches. From the test results, many recommendations are proposed.