In-process tool incidence identification based on temporal pyramid pooling and convolutional neural network

Adaptive drilling allows for the change of cutting parameters when drilling multi-material stacks, for example carbon fibre reinforced polymer (CFRP) combined with Al which are commonly found in the aerospace industry. This work proposes a deep learning approach to identify process incidences from signals acquired whilst drilling CFRP/Al stacks, which can achieve both high accuracy and immediate response by a single set of parameters. The influence of both sample length and frequency on the model’s classification performance is quantified and investigated. This work makes a notable contribution to improving the accuracy and speed of decision making for identifying incidences based on which changes in cutting parameters can be triggered in adaptive drilling of aerospace stacks.