Designed a CNC machine tools monitoring and control system based on wireless transmission technology and applied AG-based BP neural network optimization algorithm to the tool wear diagnosis. Collected the vibration signals in the process of tool machining, by means of time-frequency analysis method extracted the signal characteristic quantity which reflecting the tool wear condition as the neural network input sample, and completed the tool's wear condition recognition. The experiment result shows that the whole system works smoothly, its recognition result is accurate, and it is more practical.