Self-adapting decision support for interactive fault diagnosis of manufacturing systems

As the manufacturing environment involves more automation, fault diagnosis of automatic production systems becomes an increasingly critical issue in production control. System fault diagnosis presents considerable difficulty to the human because many diagnostic facilities available on commercial manufacturing systems are designed not for ease of fault diagnosis but for convenience of automatic control. To reduce significant production delay and expense associated with handling of system fault events, a diagnostic decision support system is highly desirable. With an understanding of specific requirements for diagnostic decision support, this paper describes a design model and relevant techniques that enable the implementation of a diagnostic decision support system (DDSS) to meet the requirements. Specifically, the DDSS is composed of a soft testing routine for selectively sampling data, an expert system for directly recognizing faults and maintaining a profile of fault occurrence and diagnosis history, a neural network for classifying faults, a fault modelling and reasoning algorithm for locating fault areas, a group of hybrid intelligent systems, expert systems, and algorithms for identifying fault components, an expert system for diagnosis explanation and verification, and a human–computer interface for exchange of diagnostic knowledge, data, process and result.