Milling cutter wear prediction method under variable working conditions based on LRCN

Milling tool wear prediction is of great significance to ensure machining quality and save production cost. Wear is a complex nonlinear physical process under the action of high pressure, high strain, and high temperature, and has different laws under different working conditions, which makes it difficult to predict tool wear under variable working conditions. In order to solve the above problems, this paper proposes a milling tool wear prediction method under variable working conditions based on spatio-temporal feature learning. Firstly, we analyze and model the working conditions affecting tool wear and pre-process the raw data to obtain high-quality data. Then, the time series raw data of milling force is converted into time series force maps, which characterize the variation of the machining process. Finally, a long-term recurrent convolution network (LRCN)-based milling tool wear prediction model is established to extract the spatio-temporal characteristics of the milling force under variable working conditions. The time series force maps and working conditions are used as input to the model. This method combines the advantages of convolutional neural network (CNN) in processing image-like data and long short-term memory network (LSTM) in processing temporal data to extract the spatio-temporal features of the time series force maps. Finally, experiments are designed to validate the method in this paper, and the results show that the method can achieve accurate prediction of the wear state of milling tools under variable working conditions.