Robust tool wear monitoring system development by sensors and feature fusion

Abstract

This study introduces a tool wear monitoring system that uses multiple sensors and a feature fusion technique. To improve the robustness of the system, different tightening torque and spindle speed conditions were considered in the experimental design stage. Vibration signals in three coordinates and sound signals were collected and transformed by fast Fourier transform for feature extraction. Two types of features in the frequency domain were used to establish the proposed system, including the mean value features selected by the class mean scatter feature selection criterion and statistical features by Gradient Class Activation Mapping++ (Grad-CAM++). The proposed monitoring system was established using a hierarchical neural network structure and feature fusion of the sensors. Cross-validation was introduced to demonstrate the performance and effectiveness of our approach under various tightening torque values and spindle speeds.

Low-Cost Real-Time Automated Optical Inspection Using Deep Learning and Attention Map

“…The attention map obtained using Grad-CAM increases the transparency of the CNN model. Although Grad-CAM provides the visualization of CNN for classification, it cannot accurately locate objects when the input images contain multiple objects in the same class [32,33]. Therefore, to identify object locations more accurately, the weight formulation connected to the feature map should be modified.…”

“…However, the pixel position is critical for an input image containing multiple objects in the same class. In Grad-CAM++, the pixel locations are considered in the weights w c k [32,33], and the calculation formula is…”

“…In this study, VGG16 with transfer learning was used to establish the proposed defect detection system. To verify the detection results, two visualization methods with attention mapping, gradient-weighted class activation mapping (Grad-CAM) [30,31], and Grad-CAM++ [32,33] were used to [34][35][36], Grad-CAM does not require model structure changes to provide effective explanations. The areas of interest of the model are denoted in the attention maps with weights.…”

SoC-Based Early Failure Detection System Using Deep Learning for Tool Wear

“…Kuntoglu et al [1] summarizes the results of the sensor and signal processing systems in the machining process. However, to ensure robustness and accuracy, sensor fusion techniques have been widely used for tool-wear monitoring [9], [10]. A robust tool wear monitoring system was established by experimental design, data collection, sensors, and feature fusion [9], [10].…”

“…However, to ensure robustness and accuracy, sensor fusion techniques have been widely used for tool-wear monitoring [9], [10]. A robust tool wear monitoring system was established by experimental design, data collection, sensors, and feature fusion [9], [10]. Subsequently, an influential sensor selection analysis was proposed to preserve the estimation accuracy and minimize the number of sensors [11].…”