Cutting tool prognostics enabled by hybrid CNN-LSTM with transfer learning

An effective strategy to predict the remaining useful life (RUL) of a cutting tool could maximise utilisation, optimise machining cost, and improve quality. In this paper, novel approach, which is enabled by hybrid CNN-LSTM (convolutional neural network-long short-term memory network) model with an embedded transfer learning mechanism, designed for predicting RUL tool. The innovative characteristics approach are that volume datasets required training deep alleviated introducing accuracy prediction. specific, takes multimodal data as input, leverages pre-trained ResNet-18 CNN extract features from visual inspection images tool, maximum mean discrepancy (MMD)-based adapt trained LSTM conduct prediction based on image aggregated process parameters (MPPs). performance evaluated in terms root square error (RMS) absolute (MAE). results indicate suitability accurate wear tools, enabling adaptive prognostics health management (PHM) tools.