Research on Dynamic Scheduling Optimization for Dongguan Smart Logistics Resources Using Reinforcement Learning

Smart logistics is an important direction in the development of modern logistics, and the dynamic scheduling optimization problem within it has become more and more a key challenge to improve logistics efficiency and resource utilization. Traditional optimization methods, such as genetic algorithms, show obvious limitations when faced with complex dynamic scenarios, and are difficult to cope with real-time changes in logistics demand and various resource constraints. To solve this problem, this paper proposes a dynamic scheduling optimization model based on Deep Reinforcement Learning (DQN), which utilizes the adaptive capability of reinforcement learning and the feature extraction capability of deep learning to provide an efficient scheduling strategy for complex logistics scenarios. During the research process, the authors constructed a simulation environment based on logistics and distribution data, and designed and verified the performance of the DQN model, taking into account the actual needs of local logistics enterprises in Dongguan. The experimental results show that compared with the traditional genetic algorithm, DQN has significant advantages in three core indicators: average distribution time, vehicle utilization rate, and unit distribution cost. Among them, the delivery time is reduced by about 20%, the vehicle utilization rate is improved to 87.3%, and the unit delivery cost is reduced to RMB 5. Further experimental analysis also shows that DQN exhibits strong stability and dynamic adaptability in the face of complex scenarios with large demand fluctuations and uneven resource distribution. The innovation of this paper is that the deep reinforcement learning algorithm is applied to the dynamic scheduling problem of smart logistics resources for the first time, which not only provides a new idea for solving the complex dynamic optimization problem, but also puts forward an operational proposal for building a smart logistics scheduling system.