Abstract: In view of the constraints caused by the special channel environment of coal mine on the target tracking of wireless sensor network (WSN) and the accuracy of measurement data, this paper designs the network topology and distributed clustering target suitable for the characteristics of underground roadway. Tracking algorithm, and based on this, the local anomaly factor detection algorithm (LOF) is used to monitor and update the wild value points existing in the measured data. Finally, the interactive multi-model filtering algorithm (IMM) is used to achieve the target state estimation. Simulation results analysis The effective balance of the algorithm reduces the network energy consumption and improves the tracking accuracy.