In the paper,a link reliability prediction model based on flexible routing in wireless sensor network is proposed.Firstly,based on the research of link reliability prediction,the link reliability prediction mapping function is established.Secondly,a supervised deep learning model is designed to predict link reliability,so as to improve the generalization ability of the model.Finally,an elastic routing model based on the current shortest path and link reliability metrics is proposed to eliminate the impact of network attacks and restore network connectivity as much as possible.In the experiment phase,compared with the link reliability prediction models such as CN,LBCN,SRW and WL,the proposed SDLM link prediction model has the best comprehensive performance,with AUC index of 0.986 4 and accuracy of 0.878 5.Compared with FRR and SQR,the proposed elastic routing model has the best comprehensive performance in transmission delay,system residual energy,average route length and node failure number.