New intelligent optimisation systems for job shop scheduling problems in the manufacturing industry

Abstract

This research focuses on scheduling systems in manufacturing and developing new optimisation techniques which are capable of dealing with scheduling problems. Scheduling is an important factor in manufacturing systems and aims to optimise the production system, reducing time and energy consumption. In this regard, numerous researchers have studied Job shop Scheduling Problems (JSSPs). In JSSPs, there are several jobs and machines, and depending on the type of job shops, schedules and related penalties, each job needs to be executed on machines on specific orders. Finding the best schedule is challenging, and there is still a need to improve and develop advanced and optimised scheduling models. This work designs optimisation models based on hybridising Genetic Algorithm (GA) techniques and Reinforcement Learning (RL) for scheduling a furnace model and simulated job shops. Hence, several sophisticated algorithms are developed for this proposal, namely Stochastic GA, Sexual GA, Ageing GA, Parthenogenetic algorithm and Ethnic GA. These algorithms are employed to establish a new metaheuristic hybrid parthenogenetic algorithm (NMHPGA) based on the combinations of the different selections to hybridise the basic GAs; moreover, two types of advanced RL, including off-policy Qlearning and on-policy RL based on State-Action-Reward-State-Action (SARSA) are developed. Following that, all algorithms are tested on two categories of scheduling job shops, including 10 single-machine job shops and 19 multi-machine job shops; all the job shops are simulated in MATLAB, and the aim is to reduce the makespan of the job shops. Results which are compared to basic GA, show that the developed models attain superior results with a faster convergence rate. As a case study, a reheating furnace model is used to optimise material heating schedules, finding the most efficient schedule to minimise time and energy consumption. The models improve the efficiency on average to 40 % on job shops and furnace fuel consumption by up to 3.20 % and operation time by 3.79 %.