Resilience-oriented integrated optimisation of resource allocation and task scheduling in complex manufacturing networks under risk propagation

Uncertainties in manufacturing usually bring about the risk of task failure, which becomes more serious for complex manufacturing service networks (MSN) by cascading effects owing to the complex connectivity and strong interdependence among resources. Hence, the propagation of failure risk throughout the MSN has to be considered in task planning. Typically, resource failures unavoidably lead to the dynamic changes in available capacity, further affect their neighbors, and the effect may propagate throughout the network. So far, fewer investigations on task planning have considered the failure risk propagation among resources. This paper addresses the resource allocation and task scheduling (RATS) considering the propagation of failure risk driven by capacity deterioration/recovery in cloud manufacturing, a typical form of MSNs. A complex-network-based risk propagation and evolution model is first developed to simulate capacity evolution with cascading effects, in identifying and forecasting the risk status of task failure. Then, a resilience-oriented RATS approach is proposed, incorporating flexible scheduling and resilience-enhanced resource grouping to disperse and isolate risk. Three simulation-derived indicators are designed to capture status of risk. A hybrid scatter search algorithm with problem-specific schemes is proposed. Extensive computational experiments and a real-world case have validated the effectiveness and practicality of the proposed method.