Deep transfer learning based diagnosis for machining process lifecycle

Faults during machining processes generate negative impacts on productivity, product quality and scrap rate. In recent years, the research of leveraging deep learning algorithms for developing fault diagnostics approaches has been actively conducted. However, the approaches have not been widely adopted by industries yet due to their inadaptability in addressing varying working conditions throughout machining process lifecycles. To overcome the limitation, this paper presents a novel deep transfer learning enabled adaptive diagnostics approach. In the approach, firstly, a Convolutional Neural Network (CNN) is designed to perform diagnostics on machining processes. Then, a transfer learning strategy is incorporated into the CNN to enhance the approach's adaptability for different machining conditions via the following steps: (1) Input datasets from machining conditions are optimally aligned to facilitate cross-domain data reuse; and (2) Weights of the trained CNN are regularized to minimize feature distribution mismatches to implement domain transfer learning. Based on the steps, the CNN can be adaptively applied across the conditions, and thereby re-training processes for the CNN from scratch can be alleviated. The developed approach was validated and benchmarked based on different parameters and settings. In the experiments, comparative results indicate that the approach achieved 94% in accuracy, which was significantly higher than other approaches without transfer learning mechanisms. Peer-review under responsibility of the scientific committee of the 27th CIRP Life Cycle Engineering (LCE) Conference.