A robust and adaptive fuzzy logic based differential evolution algorithm using population diversity tuning for multi-objective optimization. (June 2021)

• Record Type: Journal Article
• Authors: S., Brindha; S., Miruna Joe Amali
• Is Part Of: Engineering applications of artificial intelligence. Volume 102(2021)
• Publication Date: 2021-06
• Subjects: Multi-objective optimization -- Adaptive differential evolution -- Population diversity -- Fuzzy system; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2021.104240
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: This article presents an improved Multi-objective Differential Evolution based algorithm to solve multi-objective optimization problems. In the proposed algorithm named as Fuzzy Adaptive Multi-objective Differential Evolution with Diversity Control (FAMDE-DC), fuzzy system is used to control population diversity at decision variable space by self-adapting the crossover rate control parameter at various stages of evolution. Techniques such as non-dominated sorting, controlled elitism and dynamic crowding distance is used for selecting potential individuals. This control parameter adaptation and improved selection procedure results in controlling population diversity in decision space and identifying potential candidates in objective space, attaining true Pareto-optimal front with better convergence and diversity metrics. These properties make it robust and to be applied to varied problem domains without manual fine-tuning of parameters. The performance of FAMDE-DC algorithm is analysed using a set of benchmark test functions DTLZ and CEC2009 problems. Further the results are compared with other popular evolutionary based multi-objective algorithms. FAMDE-DC had a better Inverted Generational Distance (IGD) measure towards true Pareto-optimal front. The outcome of FAMDE-DC is also validated through nonparametric statistical tests Friedman and Wilcoxon signed rank test. Highlights: Self adaptation of control parameter and trial vector generation strategies. Population diversity control to enhance exploration using fuzzy logic. Improved non-domination based selection methods. Applicable to varied problem domain without manual fine tuning. Improves both decision and objective space metrics.