Solving the paintshop scheduling problem with memetic algorithms

Abstract

Finding efficient production schedules for automotive paint shops is a challenging task and several paint shop problem variants have been investigated in the past. In this work we focus on a recently introduced real-life paint shop scheduling problem appearing in the automotive supply industry where car parts, which need to be painted, are placed upon carrier devices. These carriers are placed on a conveyor belt and moved into painting cabins, where robots apply the paint. The aim is to find an optimized production schedule for the painting of car parts. In this paper, we propose a memetic algorithm to solve this problem. An initial population is generated, followed by the constant evolution of generations. Selection, crossover, mutation, and local improvement operators are applied in each generation. We design three novel crossover operators that consider problem-specific knowledge. Finally, we carefully configure our algorithm, including automated and manual parameter tuning. Using a set of available real-life benchmark instances from the literature, we perform an extensive experimental evaluation of our algorithm. The experimental results show that our memetic algorithm yields competitive results for small- and medium-sized instances and is able to set new upper bounds for some of the problem instances.