Applying a case-based reasoning method for fault diagnosis during maintenance

Abstract

The development of a fault-diagnosis system (FDS) could be helpful in identifying the source of faults before the occurrence of complete system failure. Here we report on such a diagnostic procedure for narrowing down maintenance tasks and shortening system downtime. A web-based FDS is developed based on case-based reasoning (CBR) from previous maintenance experience, for an injection moulding machine (IMM). The techniques of fault tree analysis (FTA) and information flow analysis are introduced to systematically clarify possible faults and symptoms shown by a system. A logical process is introduced to determine the correlation between the possible faults and the symptoms for computing case similarity while progressing CBR diagnosing. Some frontpages are developed for the CBR diagnosis using an ASP program in cooperation with a Microsoft Access database. The resultant FDS achieves good results with the potential for supporting remote IMM maintenance.