An adaptive neutrosophic large neighborhood search and its application

Abstract

An adaptive large neighborhood search (ALNS) lacks a clear measure for 
assessing the improvement degree in the new solution, which causes fuzziness and 
uncertainty in the operator score. To solve the aforementioned problem, the main 
innovation of this study is to propose an adaptive neutrosophic large neighborhood search 
(ANLNS). Specifically, the main work is as follows. Firstly, the number of times each 
operator scores are quantified by constructing NSs, thereby analyzing algorithm 
performance and preventing the idealization of scores. Secondly, a novel neutrosophic 
utility function and score function are proposed to assign an appropriate score for the 
operator within a reasonable interval. Finally, the effectiveness and robustness of the 
proposed ANLNS is validated by the capacitated vehicle routing problem benchmarks with 
three varying scales and comparative analyses. The compared results indicate that the 
proportion of best solutions for ANLNS are 50%, 100%, and 37.5%, which significantly 
highlight the robustness and reliability of the proposed algorithm when the degree of 
destruction is 0.3, 0.5, and 0.7, respectively. Meanwhile, the proposed ANLNS is efficient 
and flexible, providing a novel method for addressing other situation optimization 
problems.

DOI: 10.5281/zenodo.13160877