Fuzzy self-tuning differential evolution for optimal product line design

Abstract

Designing a successful product line is a critical decision for a firm to stay competitive. By offering a line of products, the manufacturer can maximize profits or market share through satisfying more consumers than a single product would. The optimal Product Line Design (PLD) problem is classified as NP-hard. This paper proposes a Fuzzy Self-Tuning Differential Evolution (FSTDE) for PLD, which exploits Fuzzy Logic to automatically calculate the parameters independently for each solution during the optimization, thus resulting to a settings-free version of DE. The proposed method is compared to the most successful mutation strategies of the algorithm as well as previous approaches to the PLD problem, like Genetic Algorithm and Simulated Annealing, using both actual and artificial data of consumer preferences. The comparison results demonstrate that FSTDE is an attractive alternative approach to the PLD problem.