In wireless sensor networks（ WSNs）

missing of sensing value is inevitable because of complex work environment.A missing value estimation algorithm is proposed based on BP neural network（ NN） model

using correlativity characteristics between multi-parameters of the same node

using strong correlative parameters set as input to carryout output estimation of missing value.In order to improve accuracy and stability of value estimation

propose combine linear regression algorithm with above algorithm

estimated values of two kinds of missing value are weighted average

effective estimations of missing value is carried out aiming at complex changing rule.Based on real sampling value

simulations and analysis are carried out

results show that the proposed algorithm can effectively achieve estimation of missing value

meanwhile the algorithm has less dependence on topological structure and node coverage rate of WSNs

so it has good practicability.