Multi-Objective Optimization of Customer-Centered Intermodal Freight Routing Problem Based on the Combination of DRSA and NSGA-III

Chunjiao Shao, Haiyan Wang, Meng YuSustainability 2022 · DOI

TL;DR

This paper tackles the complex challenge of optimizing intermodal freight routing by developing a customer-centric approach. It introduces a novel interactive method that combines the NSGA-III evolutionary algorithm with the Dominance-based Rough Set Approach (DRSA) to help shippers find the most satisfactory transportation routes by actively incorporating their preferences throughout the decision-making process.

What Problem Does This Paper Address?

The efficient routing of freight through intermodal transportation networks is crucial for global trade. However, existing approaches often focus on the operator's perspective, neglecting the diverse and evolving needs of shippers. Shippers have multifaceted requirements beyond just cost and speed, including reliability and flexibility, which are often not adequately modeled or considered in traditional optimization methods. This leads to suboptimal solutions for shippers and can hinder the adoption of intermodal transport. This paper aims to bridge this gap by creating a system that empowers shippers to actively participate in routing decisions, ensuring that the chosen routes truly align with their preferences and operational realities.

Methods

The core of this research lies in its interactive multi-objective optimization (IMO) approach. The authors propose a novel framework that integrates two key components:

  • NSGA-III (Non-dominated Sorting Genetic Algorithm III): This is a state-of-the-art evolutionary algorithm used to generate a diverse set of non-dominated solutions (the Pareto front) for the multi-objective optimization problem. It's chosen for its ability to handle multiple objectives effectively and maintain solution diversity.
  • DRSA (Dominance-based Rough Set Approach): This method is used to model the shipper's (Decision Maker's - DM) preferences. Instead of requiring the DM to assign weights to objectives upfront (a difficult task), DRSA allows the DM to classify solutions as "better" or "otherwise." Based on these classifications, DRSA derives interpretable "if-then" decision rules that capture the DM's preferences. These rules are then translated into constraints for the NSGA-III algorithm in subsequent iterations, guiding the search towards more satisfactory solutions.

This iterative process allows the DM to learn about their preferences and the trade-offs involved, leading to a more informed and satisfactory final decision.

Datasets

The study utilizes a case study based on the import needs of a shipyard in Shanghai, China, from Italy. The origin and destination were set as Genoa and Shanghai, respectively. Five freight batches were considered, with data including release times, permissible delivery time ranges, and ideal delivery time ranges. The transportation network was constructed with 19 nodes and 312 inter-node transportation services, incorporating data on international rail, sea, and air transport, as well as road and inland waterway connections. Specific data on service frequency and latest booking times were also collected.

Key Insights and Ideas

  • Customer-Centricity is Key: The paper strongly emphasizes that optimizing for shipper preferences, not just operator costs, is crucial for the practical success of intermodal freight routing.
  • Interactive Preference Elicitation: The combination of NSGA-III and DRSA offers a powerful way to elicit complex shipper preferences without overwhelming them. The "glass box" nature of DRSA makes the decision rules interpretable, fostering trust and understanding.
  • Trade-offs are Dynamic: The case study demonstrates that shippers' preferences can evolve during the decision process. The interactive approach allows for adjustments and refinements, reflecting real-world decision-making.
  • Service Type Matters: Different transportation modes (air, sea, rail) exhibit distinct trade-offs between cost, timeliness, reliability, and flexibility, influencing the optimal choice for different shipper requirements.
  • Reliability and Flexibility are Linked: The study found a positive correlation between reliability and flexibility objectives, while cost generally showed a negative relationship with these two.

Key Results

The proposed interactive method was applied to five freight batches, demonstrating its effectiveness in finding satisfactory solutions. For Batch 3, the interactive process led to the selection of a rail-international rail-track route that balanced cost and reliability, with the DM iteratively refining their preferences. The study also compared the interactive method with priori (weight-based) and posteriori approaches, showing that the interactive method allowed for more nuanced preference expression and potentially more tailored solutions. For instance, for Batch 3, the interactive method yielded a more reliable solution compared to a posterior method, even if slightly more costly, reflecting the DM's adjusted priorities.

Limitations and Future Work

The authors acknowledge several limitations:

  • Reliability and Flexibility Modeling: The current models for reliability and flexibility are somewhat simplified. Future work could incorporate a broader range of factors, such as network disruptions for reliability and changes in demand volume for flexibility.
  • DM Selection and Performance Evaluation: The study focused on a single expert DM. Future research could explore group decision-making and conduct more extensive user experience validation to assess the interactive approach's performance and user participation benefits.

The paper suggests that future efforts could also focus on refining the modeling of these objectives and extending the application of the interactive approach to further demonstrate its value.

Summary updated

Abstract

The satisfaction of requirements and preferences of shippers is critical to enable the practicability of solutions that are derived from intermodal transportation routing problems. This study aims to propose a decision process to help shippers participate better in routing decisions. First, we considered shippers’ requests on transportation cost, timeliness, reliability, and flexibility to construct a multi-objective optimization model. Then, to solve the interactive optimization method that was proposed, NSGA-III was applied to obtain the Pareto front and dominance-based rough set approach to model the preference information. Finally, a case study was conducted and an expert was invited as decision-maker to demonstrate the applicability of the proposed model and the effectiveness of the interactive method for shippers. The results are expected to provide shippers with more rational transportation schemes and insights for the sustainable development of intermodal transportation.
BibTeX
@inproceedings{shao2022multi,
  author = {Chunjiao Shao and Haiyan Wang and Meng Yu},
  title = {{Multi-Objective Optimization of Customer-Centered Intermodal Freight Routing Problem Based on the Combination of DRSA and NSGA-III}},
  booktitle = {Sustainability},
  year = {2022},
  doi = {10.3390/su14052985},
}
Figure