To avoid the pre-designed " teaching" or " training" phase for the condition monitoring system at present,an intelligent monitoring system for high-speed milling process is proposed based on the analysis of the tool wear rule and different tool wear stages.Self-learning was introduced to the system to automatically identify different tool wear states and estimate the wear value,which is independent of the pre-designed " teaching" or " training" phase.Three-direction components of the cutting force signals generated in high-speed milling process were processed using discrete wavelet decomposition technology.Features in different time and frequency domains were extracted and selected through correlation analysis method.The real-time intelligent monitoring system was built on the cycle process of linear fitting and Mahalanobis distance(MD) calculation.A series of experiments on a CNC vertical milling machine tool shows that the proposed method is accurate for feature extraction and efficient for condition monitoring of cutting tools.