Abstract: Aiming at the problem of energy constraints in data transmission, to reduce the amount of sensory data communication and energy consumption in wireless sensor networks, we propose an effective data fusion strategy based on the regression function of sensory data. This strategy adopts the weighted least squares method and the kernel function method to optimize the sensory data model. Rather than transmitting all measurement data to nodes, nodes transfer the model parameters only. Experiment results show that the proposed data transmission strategy, which estimates induction data in the monitoring area, can reduce the total energy consumption and prolong the lifetime of wireless sensor networks, and it is effective in practical application.