Comparison of the Scalarization Approaches in Many-Objective Simulation-Based Optimization in Production System Control

摘要

Complex manufacturing systems control, involving contradictory production goals, is a challenge in spite of many multi-objective optimization decision-making strategies. In this article we demonstrate an application of the scalarization approach with a priori articulated preferences in the multi-objective optimization via discrete-event simulation in the field of the manufacturing product control. Especially, we focus on the effect of the used scalarizing multi-criteria function form combining all considering production goals using user-supplied weights. Four key performance indicators: costs per part unit, an average flow time, machines utilization and an amount of production for studied production system were finding by simulation in simulator Witness and optimized simultaneously under specific constraints. Metaheuristics like Thermoadaptive Simulated Annealing and Random Solutions and at last a brute force algorithm were used as optimization methods for finding global minimum of scalar functions generated by different methods of scalarization.