Production Scheduling for Energy Cost Optimization with Constraints: Comparison of Genetic Algorithm and PSO Techniques

Intelligent production scheduling has become predominant in smart factories that aim to greatly reduce energy costs. For this purpose, scheduling systems need to not only consider multiple machines, PV energy, and electricity price volatility but also take into account constraint applicability and flexible job shop layouts to meet smart factories’ demands. To solve these complex scheduling problems, Artificial Intelligence (AI) is the ideal approach. Accordingly, the premise of the present paper is to implement and compare quantitatively three AI algorithms for intelligent production scheduling: a proposed Genetic Algorithm (GA), and a conventional GA and Particle Swarm Optimization (PSO) from the DEAP Python library. Different manufacturing concepts are considered to solve the aforementioned issues, including tasks, machines, products, and manufacturing orders. In addition, three constraints are explored: schedule completion deadline, manufacturing order deadline, and time leap. Validation of the AI algorithms for production scheduling is done through a real case study with real-production data. Results show that the proposed GA has, on average, lower energy costs of about 20.2% and 22.2% when compared to the conventional GA and PSO, respectively. Furthermore, when constraints were imposed, only the proposed GA was effective.