A two-stage fault tolerance method for large-scale manufacturing network

Nowadays, networked manufacturing paradigm, such as cloud manufacturing, has brought new challenges for fault tolerance methods. Failures and errors of manufacturing services are unavoidable in the large-scale network. Appropriate fault tolerance methods need to be adopted to improve the reliability of the whole manufacturing network. Aiming at solving this problem, a two-stage fault tolerance method is, thus, proposed. In the off-line stage, a LeaderRank-based manufacturing nodes ranking (LNR) algorithm is put forward to rank the manufacturing services according to their significance in fault tolerance. Replication strategies are employed only for critical services to achieve the tradeoff between fault tolerance effect and fault tolerance loss. In the online stage, an A∗-based heuristic alternative path searching (AHPS) algorithm is proposed to find suitable replacement schemes for composite services. The experimental results illustrate that the two-stage fault tolerance method can improve the reliability of large-scale manufacturing networks both effectively and efficiently.