Meta‐heuristics for real time routing selection in Flexible Manufacturing Systems (FMS)

Abstract

Most studies in real-time Flexible Manufacturing System (FMS) scheduling and control area do not consider the effect of routing flexibility; their focus is typically on use of scheduling (i.e., dispatching) rules based on routing selection carried out prior to production. Such an approach is not applicable to random-type FMS, in which no knowledge about incoming part types is available prior to production. For such a scenario, parts can have alternative routings, even for parts of the same type. Thus, the control system of a random-type FMS requires the capability to adapt to the randomness in arrivals and other unexpected events in the system by effectively using operation and routing flexibility in real-time. In this chapter, the objective is to present a comparative study of a group of meta-heuristics, including tabu search (TS), ant colony optimization (ACO), genetic algorithms (GA), particle swarm optimization (PSO), electromagnetic meta-heuristic (EM), and simulated annealing (SA), against the Modified Dissimilarity Maximization Method (Modified DMM). DMM (Saygin and Kilic 1999) is an alternative process plan selection method originally proposed for the routing selection in off-line scheduling of an FMS. In subsequent studies (Saygin, Chen, and Singh, 2001) and (Saygin, Chen, and Singh, 2004) DMM has been: (i) used as a real-time decision- making tool to select routings for the parts that are in the system, (ii) tested and benchmarked against First-in-First-out/First Available (FIFO/FA) and Equal Probability Loading (EPL). Based on the DMM model, a modified DMM (Hassam and Sari 2007) is developed for selection of alternative routings in real time in an FMS. Modified DMM method improves the performances of the FMS in terms of higher production rate, higher utilization rate of the machines and the material handling system.