Multi-objective optimization for multimodal transportation routing problem with stochastic transportation time based on data-driven approaches

We study a multi-objective optimization model of a stochastic multimodal transportation network considering key impact factors such as transit cost, time, and transport mode schedule while minimizing total transportation cost and transportation time. In this study, we apply the Monte Carlo simulation to deal with the stochastic transportation time in the network and propose a data-driven approach that combines historical data and the dataset generated by the data mining algorithm to accelerate the search for the nondominated solution in the simulation. To validate the effectiveness of the proposed Data-Driven Multi-Objective Simulation Ant Colony (DD-MSAC) algorithm, we compare the optimum-seeking performance and the running time consumption of the Nondominated Sorting Genetic Algorithm-II (NSGA-II) and the Multi-Objective Simulation Ant Colony (MSAC) algorithm. Then, the MSAC algorithm is adopted as the benchmark for the comparison study on the solving performance of the proposed DD-MSAC algorithm. We conducted 30 times simulation run under different network scales in our numerical examples to show that the DD-MSAC algorithm can be equally effective as the non-data-driven MSAC algorithm in finding a nondominated solution as the average error does not exceed 5%. Meanwhile, we analyze the impact of different data-driven approaches, including data pool and support vector machine, on the solution quality and the running time. Finally, we use an example of China’s Belt Road Initiative to verify the effectiveness of the proposed algorithm.