Aiming at the problem of on-line fault monitoring of automatic machine tools,a monitoring system based on linear separable SVM was designed.The hardware structure of the monitoring system included STM32F103ZET6 single chip microcomputer,sensor module,memory module,communication module and display module.In the software algorithm flow,the linear separable SVM classifier could ensure the maximum geometric interval from the data set to the optimal hyperplane,improve the reliability of the data points closest to the optimal hyperplane and the classification accuracy of fault samples.The test results show that the proposed system has fast data training convergence speed,high fault data classification accuracy and performance advantages over the traditional monitoring system.