An algorithm for constructing a failure propagation tree in manufacturing systems

The authors consider the diagnostic step of the error recovery process in manufacturing systems and formalize an algorithm for constructing and pruning a failure propagation tree. They also present a methodology for classifying error reasons in order to reduce the diagnostic search space. This algorithm assumes the availability of a hierarchical process plan for the manufacturing operations. The process plan is expanded into a hierarchical operator model. The algorithm provides an expansion rule and a priority rule. The expansion rule prescribes the manner in which operations are traced back to previous operations at the same or lower levels through precondition nodes and at higher levels through dummy nodes. The priority rule determines the order in which the tracing should occur. Procedures for pruning the failure propagation tree are presented as an expanded algorithm. The methodology cannot only locate the possible failure causes and failure sources in the failure propagation tree but it lays the groundwork for increasing machine reasoning for failure analysis and for automated error recovery. An illustrative example is presented