Optical fiber allocation in cable manufacturing is a core process for reducing material costs, improving production efficiency, and optimizing inventory management. However, factors such as variations in optical fiber lengths and storage shelf-life significantly increase the complexity of the problem, making traditional heuristic algorithms difficult to apply. To address this challenge, this paper proposes a modified proximal policy optimization (MPPO) algorithm. First, a Markov decision process (MDP) is constructed for optical fiber allocation, designing dynamic state and action spaces along with the corresponding reward function. Second, an exponential decay function is introduced to design an adaptive clipping coefficient, thereby improving the constraint on the divergence between the new and old policies in the proximal policy optimization (PPO) algorithm. Finally, the network model of the PPO algorithm is enhanced by adding hidden layers and incorporating a previous actor network, thereby improving the agent’s flexibility and perceptual capabilities. The results show that the proposed method exhibits superior performance and stability, highlighting its potential in optical fiber allocation.
原文链接
分享
引用
