The Study On Intelligent Detecting Technology Of Ball-End Mill Wear

With the development of mechanical automation, more and more attention are paid to on-line monitoring of processing. In the machining process, tools will be abraded continuously, even damaged suddenly. The traditional monitoring method is parking first and unloading the tools, then measuring geometric parameters of tools. This method increases parking times and reduces the machining efficiency, so it is necessary to monitor the wear of tools on line .The core of the entire monitoring system is measuring technology which can measure cutting forces, electric current, sound, power and so on. Because of easy attained and fast response, cutting forces are paid more attention. The paper carrys on cutting forces combined with intelligence methods to study the relevant mesuring techonology.In the paper, the object of research is the process of milling brittle materials(quartz galss) by carbide ball end milling. All the experiments are done on the milling experimental platform which includes several parts, such as spindle, clamps, force measurement, control and shock absorption part, etc. The platform can move for three vertical directions by controling the electrical machines. In the process, the data of milling force can be acquired through dynamometer.The signal processing technology includes time domain, frequency domain and time-frequency analysis. The acquired forces data are analysed by time domain analysis method firstly. Then through the methods of frequency domain and time-frequency analysis, some features about different wear states of ball end milling are concluded. The different states are early wear, normal wear and sharp wear.The intelligent method used in the paper is Back Propagation neural network which is made up of input layer, hidden layer and output layer. The features reacting tools'different wear states can be used as input layer of BP neural network, the transfer function of hidden layer is sigmoid function(logsig), and the output layer is three-state of ball end milling. An ideal BP neural network model has been gotten finally through sample training and valve threshold adjustment, and this model can forecast tools'states precisely.