Research on WSN Optimization Strategy Based on Improved DDPG Algorithm
Deep Deterministic Policy Gradient(DDPG)is a combination of Actor-Critic and DQN algorithms.As one of the most classic algorithms in deep reinforcement learning,DDPG algorithm is widely used in wireless sensor networks(WSN).A weighted information entropy depth deterministic policy gradient(WIE-DDPG)algorithm is proposed to solve the problems of DDPG algorithm such as low training efficiency,slow convergence speed and large synchronization error.The algorithm assigns weights to the training data in advance,trains the data according to the weight ratio,and integrates the results through neural networks.The experimental results show that compared with DQN and DDPG algorithms,WEI-DDPG algorithm has higher training efficiency,faster convergence speed and lower synchronization error.
国家科技期刊平台
NSTL
万方数据
