Optimizing cruising speed for energy-efficient and safe operation of metro-based underground logistics systems using a surrogate-assisted multi-objective evolutionary algorithm

The metro-based underground logistics system (ULS) offers an effective solution to alleviate traffic congestion, environmental degradation, and energy inefficiency caused by intensive road-based freight transport. Unlike passengers, cargo exhibits variable mass properties, significantly influencing the dynamic performance of metro vehicles. To enhance both operation safety and energy efficiency of metro-based ULS, this study proposes an AI-driven optimization framework for optimizing the cruising speed of metro vehicles using multibody dynamics (MBD), Random Forest (RF) surrogate modeling, and adaptive reference point-based multi-objective evolutionary algorithm (AR-MOEA) that establishes a complete intelligent decision pipeline. MBD simulations under 5625 designed working conditions generate training data covering comprehensive working conditions. An RF-based surrogate model learns the complex vehicle dynamics to enable rapid performance prediction. AR-MOEA optimizes cruising speeds across segments in 4.9 s to simultaneously improve safety and energy efficiency. Compared to a uniform speed profile, the proposed AI-powered strategy can improve the operation safety by up to 23.51 % and energy efficiency by up to 20.75 %. This study provides valuable insights for the sustainable development of metro-based ULS.