If the fault is found in planetary gear of cutting unit in the shearer, the shearer will not be successful mining. In this paper, first, a variety of failure signal of reducer is collected and the parameters of support vector machine are optimized using PSO and the optimal classifier is obtained. second the collected various signals are processed using classifier of Support Vector Machines. The results show that this method can diagnose the various fault of reducer quickly and accurately. The proposed method is contribute to achieve fault diagnosis of planetary gear real-time and automatically.