An Enhanced Artificial Bee Colony Algorithm for Flexible Job Shop Scheduling With Multiple Assembly Precedence Constraints

Abstract

This study proposes an enhanced artificial bee colony algorithm (ENABC) for flexible job shop scheduling with multiple assembly precedence constraints. The objective is to minimize the total penalty cost (TPC), defined as the sum of tardiness, earliness, flow time, and setup costs. ENABC enhances the classical artificial bee colony algorithm through five mechanisms: a three-parameter encoding scheme with efficient decoding heuristics, a dual food-source exploration strategy, a modified precedence-preserving crossover operator, a modified tabu search strategy, and a tabu list to prevent revisiting solutions. The proposed algorithm is evaluated using 10 large-scale industrial problem instances. Compared with benchmark approaches based on the TLSAPVNS and SCEGWO frameworks, ENABC achieved average TPC reductions of 12.96% and 10.90%, respectively. Relative to the factory's existing scheduling practice, the proposed approach reduced the average TPC by 49.93%. Furthermore, it required less computation time than all competitors.