Intelligent fault diagnosis of drill bit using Digital Twin Assisted deep learning classifiers of MLP, CNN, Hybrid MLP-CNN, and Hybrid CNN-MLP with PCA algorithm

Intelligent fault diagnosis using AI models is an evolving engineering technique in the manufacturing sector. It detects anomalies in machinery and improves productivity, safety, efficiency, and economy in industries. Many new models have been developed recently to improve the accuracy of predicting various machine faults. A Digital Twin (DT) is a digital proxy of physical assets, an emerging technology in intelligent manufacturing that can address the faults in drilling machines. This work introduces a dynamic model of the drill bit with wear to generate simulated drill bit vibration signals. A digital twin model is applied using LabVIEW, where the model parameters are updated on the actual behavior using a physical model. In the second step, all necessary tool conditions are simulated using the digital twin output. In the next step, feature extraction uses Principal Component Analysis (PCA) to improve classification efficiency. Deep learning techniques of MLP and CNN are applied to predict the tool faults with possible variations in working conditions and system characteristics. Then, a hybrid deep learning algorithm of MLP-CNN is also used to indicate the wear of the drill bit tool. After applying deep learning techniques with the integration of digital twins and Principal Component Analysis (PCA), a comparative analysis was done in the final stage to conclude the effectiveness of the proposed model. The experimental result says that DT with hybrid MLP-CNN is more accurate in predicting drill bit faults, with the highest accuracy of 100% with a limited amount of measured data.