Capacity Control and Simulation in Multi-Level Fare Class Based on Enhanced Exploration PPO Algorithm

The traditional cabin capacity control method has limited real-time response ability and is difficult to achieve good results in complex and changing environment. A capacity control algorithm of multi-level fare class based on enhanced exploration PPO algorithm is proposed. The problem of capacity control is transformed into a sequential decision problem. Using deep neural network fits the value function, the state space, action space and reward function for solving sequential decision problem are given. The simulation results show that compared with benchmark algorithms such as EMSR-b, FCFS and PPO, the proposed algorithm could effectively obtain more benefits.