Variable-Relationship Guided LNS for the Car Sequencing Problem

Abstract Large Neighbourhood Search (LNS) is a powerful technique that applies the “divide and conquer” principle to boost performance of solvers on large scale Combinatorial Optimization Problems. In this paper we consider one main hindrances LNS popularity, namely requirement an expert define problem specific neighborhood. We present approach learns from structure search in order generate neighbourhoods can match domain heuristics developed by expert. Furthermore, new objective function for optimzation version Car Sequencing Problem, better distinguishes solution quality. Empirical results public instances demonstrate effectiveness our against both heuristic state-of-the-art generic approaches.