In the most existing connected dominating set (CDS) based coverage algorithms, the mechanisms of maximizing sleep node numbers is adopted to save energy in WSN. Active nodes of WSN cause rapidly exhausts energy which leads to CDS failure and fade coverage. In addition, frequently activating other nodes speeds up network topology changes, and lead to the network convergence problems. A learning automata-based area coverage algorithm is proposed for WSN. d-CDS is adopted to construct network topology, and learning automata is used to select the optimal node of current sensors. Then the constructed CDS can be optimized, and the load balance of the active nodes is realized to improve network coverage performance. Finally, Simulation experiments are conducted to compare Gossip, ST-MSN and TMPO. The results show that, the proposed algorithm is superior to the three algorithms in network coverage rate and the residual energy of active nodes.