Research Of Milling Force Prediction Model Based On Feed Motor Current

During NC machining, the factors such as change in the depth of cut, material hardness variation and tool wear may result in the conservative feed rate selected and therefore insufficient use of NC machining tools. It is a important way to increase productivity by real-time monitoring and adaptive adjusting the machining process.The method that the cutting force is measured directly by the 3-D dynamometer is only used in laboratory, can't be used in industry. As a result, an indirect milling force measurement method based on feed motor current has been proposed. The dissertation focuses on the following research work.Analyzing the transfer of milling force and the process of power conversion between mechanical and electrical power particularly, the relation of the milling force and feed motor current has been proved. Using the data from the experiments to analyze the milling force signal in both time and frequency domain, extracting the value of the milling force. Analyzing the feed motor current signal in both time and frequency domain, using the wavelet analysis method to extracting the current which has the same frequency as the milling force. The average value of this transformed current is used to be the value of feed motor current in token of milling force.Because there is extraordinary non-linear relation between feed motor current and milling force, the BP arithmetic of artificial neural network method is used to establish the model of the relation. The value of feed motor current, feed speed and spindle rotate speed are used to be the input of the model, the milling force is used to be the output. After training by a lot of experimental data, optimizing the parameters of the net, a neural network model which is adapted to forecast milling force base on feed motor current has been established.At last, using the experimental platform which is established to measure the milling force and feed motor current, taking several representative experiments to validate the capability and precision of this model, and analyzing the factors which have an effect on the model. Finally the validity of the prediction model has been proved.