Abstract: For the particularity of coal mine wireless sensor network structure and the impact of nodes clock instability on synchronization precision, we proposes a time synchronization algorithm based on the optimal linear fitting. On the basis of analyzing the traditional TPSN（Timing-sync Protocol for Sensor Networks）algorithm, it establishes a new type of algorithm model, and by using linear regression algorithm compensates clock drift and filters abnormal data points which improves the accuracy of synchronization, reduces the influence of abnormal data points on the synchronization accuracy, and extends the synchronization time. The experimental results show that the proposed algorithm can effectively improve the synchronization accuracy and stability.