In wireless sensor networks attacks, aiming at the low aggregate precision of conventional method, introduce a new data aggregate method. The new method adopts perceptive data to evaluate the sensor variance and carry out weighting evaluation, which effectively reduce the node measurement error. At the same time, the new method can evaluate support rate of every cluster object variance according to index attenuation function. It can reduce the jamming of network noise. Finally, realize the high precision aggregation. The simulation results show that the new method aggregate function is better than the conventional one, which can effectively reduce the node measurement error and network noise jamming.