Strategic decision-making for sustainable production and distribution in automotive industry: a machine learning enabled dynamic multi-objective optimisation

Over the last decade, numerous researchers have disclosed that major automotive companies do not conform to regulatory or societal expectations regarding their environmental and social performances. This paper explores the dynamic capabilities of production distribution within the sustainability practices of automotive industries. It offers insights to better grasp and articulate the environmental, economic, and social dimensions of sustainable supply chains. The research framework encloses all supply chain phases, from raw material sourcing to retailing finished products. Three conflicting objective functions are identified: social advantages maximisation, cost minimisation, and emission minimisation. Specifically, the study tackles a dynamic multi-objective optimisation model where each automobile type faces a series of dynamic demands. The dynamic nature of the problem poses significant challenges to conventional evolutionary algorithms for detecting the optimal solutions over time. Therefore, we introduce an interconnected prediction-based dynamic non-dominated sorting algorithm (ICP-DNSGA-II). Finally, extensive computational experiments are conducted to assess the effectiveness of this holistic approach. The findings offer valuable insights for automotive industry stakeholders and policymakers, illustrating its potential to enhance operational efficiency and sustainability performance across the supply chain. Most importantly, this paper proposes an automated decision-making approach to generate optimal solutions with dynamic changes in market demands.