Failure evolution analysis for complex human-machine system: A case for nuclear power system

Mission failure evolution is a unique aspect of complex human-machine system (CHMS), which contains failure regularity and directs emergent reconfiguration measures. Oriented to all varieties of system components, an analysis of how the parts are connected and contribute to CHMS failure is necessary. Prior work paid more attention to system design with the law of system normal operation, where a single function failure was identified positively and its consequence was analyzed to guarantee reliability and safety. In this paper, a normally operational CHMS process is divided into three layers which contains task, element and capacity as well as the conceptions of Basic Task Space (BTS) and Repair Task Space (RTS). The measurement method of these three aspects was defined. Furthermore, CHMS failure measurement will be presented with the parameters of capacity deviation. Taking multiple elements coordination and self-organization into consideration, a multi-agent based simulation approach of system failure evolution path is proposed. This method is demonstrated with the nuclear power system intending to interpret how the system fails to deal with all possible disturbance. Obvious results show that the multiple failure paths can be obtained, from which the writer provides some constructive comments for system design.