Title: HKn-RVEA: a novel many-objective evolutionary algorithm for car side impact bar crashworthiness problem
Authors: Gaurav Dhiman; Amandeep Kaur
Addresses: Department of Computer Science, Government Bikram College of Commerce, Patiala, 147001, Punjab, India ' Department of Computer Science and Engineering, Sri Guru Granth Sahib World University, Fatehgarh Sahib, 140406, Punjab, India
Abstract: In this paper, a novel hybrid many-objective evolutionary algorithm, named as hypervolume indicator based on knee point driven and reference vector guided evolutionary algorithm (HKn-RVEA) is proposed. HKn-RVEA is based on the hypervolume indicator, knee points, and reference vector adaptation strategies. The knee points are used to improve the search ability. The reference vectors are used to decompose the optimisation problem into a number of sub-problems. In the proposed algorithm, an adaptation strategy is used to adjust the distribution of the knee points and reference vectors. The proposed algorithm is compared with five well-known evolutionary algorithms over standard benchmark test functions. The results show the performance of HKn-RVEA is better than the competitor algorithms in terms of inverted generational distance (IGD) and hypervolume (HV ) performance measures. HKn-RVEA is also applied to real-life car side crashworthiness problem to demonstrate its efficiency. The experimental results show that the proposed algorithm is able to solve many-objective real-life problems.
Keywords: many-objective optimisation; HypE; hypervolume estimation algorithm; reference vector guided evolutionary algorithm; RVEA; knee points; convergence; diversity.
International Journal of Vehicle Design, 2019 Vol.80 No.2/3/4, pp.257 - 284
Received: 10 Dec 2019
Accepted: 03 May 2020
Published online: 28 Sep 2020 *