Tugboat Dynamic Scheduling Method Based on Improved Sarsa Algorithm
Aiming at the shortcomings of the traditional Sarsa algorithm,the optimization of tugboat dynamic schedu-ling method is studied.Based on the reinforcement learning framework and the state and environment information of tugboats,the state-action function is established to search the optimal strategy of tugboats scheduling decision.The update method of Q function in Sarsa algorithm is changed to overcome the problem of slow convergence.At the same time,according to the learning rate and action selection mode,the exploration strategy and utilization strategy are balanced to improve the convergence speed and performance of the algorithm.The simulation results show that the algorithm can effectively shorten the waiting time and improve the utilization efficiency of tugboat resources.
国家科技期刊平台
NSTL
万方数据
