Two-Stage Assembly Scheduling with Batch Setup Times, Time-Dependent Deterioration, and Preventive Maintenance Activities Using Meta-Heuristic Algorithms

Abstract

This article considers a two-stage assembly scheduling problem (TSASP) with batch setup times, time-dependent deterioration, and preventive maintenance activities (PMAs). The objective of this problem is to simultaneously determine the optimal component-manufacturing sequence (CMS), product-assembly sequence (PAS), number of setups, and number and position of PMAs (PPMA). First, to determine the optimal solution, a novel mixed integer linear programming model (MILP) for the proposed problem is derived. Then, a standard genetic algorithm (SGA), hybrid genetic algorithm (HGA), standard harmony search (SHS), hybrid harmony search (HHS), and harmony-search-based evolutionary algorithm (HSEA) were proposed owing to the intractability of the optimal solution for large-scale problems. SGA and SHS provide a chromosome to represent a complete solution including three decisions (CMS, PAS, and PPMA). HGA, HHS, and HSEA provide a chromosome to represent a partial solution including PAS. CMS and PPMA are found by an effective local search heuristic based on the partial solution. A computational experiment is then conducted to evaluate the impacts of the factors on the performance of the proposed algorithms.

Production, maintenance and resource scheduling: A review

“…The studies by Yang et al (2008) , Lee & Kim (2017) , Cui, Lu, Zhou, & Han (2016 , Xing, Qiao, & Lu (2019) , Feng, Tan, Xia, Pan, & Xi (2019 , Yu & Seif (2016) and Jung, Woo, Koh, & Kim (2018) consider scheduling maintenance before reaching a certain machine age. In Yang et al (2008) , maintenance is performed before reaching a limit on the consecutive number of produced jobs.…”

“…In such case, maintenance can be started at any time unless the cumulative working time after the end of the previous maintenance exceeds a certain limit. The studies by Yu & Seif (2016) and Jung et al (2018) examine a maintenance problem where each job deteriorates the machine at a different rate. Yu & Seif (2016) consider multiple ML which are reached through the deterioration of the jobs.…”

“…Maintenance in each ML has a different cost and restores the machine to the best ML. Jung et al (2018) on the other hand consider maintenance that must be scheduled before reaching a threshold which is reached through the job degradation. Cui et al (2016) examine two variants for the age threshold: (1) idle time is included in the cumulative working time and (2) idle time is not included.…”

“…Mathematical formulations are derived and mathematical programming models are provided in all studies except for Yang et al (2008) , that instead presents polynomial solvable solutions to a set of problem instances. Also, metaheuristic solutions are provided in Cui et al (2016) , Xing et al (2019) , Feng et al (2019 , Yu & Seif (2016) and Jung et al (2018) . All studies, except for Cui et al (2016) and Yu & Seif (2016) , provide both a GA and an additional meta-heuristic solution, where both algorithms are compared.…”

“…All studies, except for Cui et al (2016) and Yu & Seif (2016) , provide both a GA and an additional meta-heuristic solution, where both algorithms are compared. The HGA by Xing et al (2019) and the Harmony Search (HS) algorithm by Jung et al (2018) prove better than the standard GA.…”