Multi-objective optimization of truck-drone cooperative routing problem based on customer classification and fuzzy time windows

With the growing demand for personalized logistics services, the combined use of drones and trucks as collaborative delivery services has become increasingly crucial for improving service levels. A key challenge lies in the rational allocation of limited logistics resources to enhance customer satisfaction. To address this challenge, this study proposes a novel multi-objective optimization model for truck-drone collaborative routing, utilizing customer value classification and fuzzy time window management. First, considering the Pareto principle (also known as the 80/20 rule) of customer profitability, customers are classified into three levels: high, medium, and low based on their current purchase value (CPV) and potential purchase value (PPV). This classification allows for a differentiated delivery strategy: high-level customers receive door-to-door delivery via drones, medium-level customers are served by trucks at designated pickup nodes, and low-level customers are directed to centralized self-pickup locations. Second, to better accommodate customer preferences, flexible time windows are introduced, including desired and tolerated time frames, with varying sensitivity coefficients assigned to different customer levels. Finally, a multi-objective optimization model is constructed to minimize costs and maximize customer satisfaction. To solve this model, a hybrid genetic algorithm-simulated annealing (GA-SA) approach is employed, incorporating dynamic adjustment strategies and a fast, non-dominated sorting algorithm to enhance computational efficiency. Benchmark instances are used to evaluate the proposed algorithm, demonstrating its capability to generate high-quality solutions. Additionally, a real-case study in Chongqing, China, validates the effectiveness of both the proposed model and algorithm. The results indicated that while the costs of truck-drone collaborative delivery were comparable whether or not customer classification was considered, customer satisfaction improved by 22.11% when classification was taken into account. This proves the potential of the proposed delivery strategy to enhance customer satisfaction while optimizing logistics delivery routes. The findings also have practical implications for various supply chains, confirming that integrating our proposed framework can significantly improve customer satisfaction.