Abstract: Aiming at problems of the existing traffic shaping algorithms in applications of sensor network, a novel traffic shaping algorithm by variable weight combination forecast based is proposed. The reasons of sensor network traffic’s sudden randomness and time-varying unequilibrium are analyzed. By modeling the network traffic of fuzziness, randomness and time-varying, based on variable weight combination forecast a Traffic Shaping Algorithm with Variable weight combination forecast is put forward. To get more accurate forecast of traffic value, by optimal combination of two prediction method, Fuzzy AR and Kalman, traffic shaping algorithm is plan ahead shaping rate in order to smooth the output packet flow. Experiments show that the traffic shaping algorithm we proposed can predict the sensor network flow precisely, reduce the packet dropped rate and increase network throughput at the same time, so it can improves the sensor networks’ information transmission performance of QOS.