Abstract: To solve the problem that there is combination problem when multi-state system reliability model is formulated by traditional methods, the multi-state reliability model of Reconfigurable manufacturing system (RMS) is built based on the stochastic process and the universal generating function (UGF) technique. In order to get all the steady-state probability, the Markov process model of reconfigurable machine tool (RMT) with the characteristic of non-adjacent state transition is established. Through defining the buffer states and analysing the influence of buffer states on the machine availability, both the buffer availability and the actual steady-state probability are obtained. According to the characteristics of RMS and RMT, UGF technique is perfected, and by which generating functions of machine level and system level are expressed followed by the establishment of multi-state reliability model of repairable RMS for multi-parts family. Furthermore, the performance indexes of RMS are also discussed. A parallel-serial hybrid machining system with four stages for double-parts family composed of cylinder heads is taken as an example to verify the effectiveness of the model and the validity of the analysis method.

Key words: Multi-state system reliability, Reconfigurable manufacturing system, Universal generating function