Tool Wear Prediction Based on Edge Data Processing and Deep Learning Model

Abstract

In order to improve the accuracy of tool wear prediction and enhance the real-time application in industrial sites, a tool wear prediction method based on edge data processing and CNN-BiGRU neural network is proposed. This method first implements data preprocessing on edge nodes, effectively reducing the amount of data transmission to avoid network link congestion. After that, the CNN-BiGRU neural network was deployed in the cloud for model training. Experimental results show that the tool wear prediction method based on edge data processing and CNN-BiGRU neural network has good real-time performance and high prediction accuracy in simulated industrial field applications.

Improving Tool Wear Prediction with Synthetic Features from Conditional Generative Adversarial Networks

“…Then linear regression is followed to learn an RUL predictor from outputs of the last LSTM unit's forward and backward hidden states. Yan et al replaced LSTMs with more recent recurrent architecture BiGRU units for RUL estimation of CNC (Computer Numerical Control) milling tool [11]. In [24], Xu et al proposed an MCGRU model where six separate CNN branches are used for feature extraction from multi-sensor readings in parallel but with different sizes of convolution kernels in each branch.…”

“…Their description is given in Table IVa. To compare the proposed methods with [11], [12], [15], [24], we used three cutter records C 1 , C 4 and C 6 , as described in Table IVb. And the average value of flank wear from three flutes was used for model training and performance evaluation.…”

Tool wear prediction method based on the SVM-Clara model

“…Tool wear can lead to reduced tool life, reduced surface quality, and increased material loss. Therefore, the wear of cutting tools plays a key role in increasing productivity [6] and improving product quality. Based on studying the rationality of the geometric parameters of cutting tools, it is necessary to observe the wear conditions of cutting tools and analyze their causes, which is very necessary for in-depth research on the wear laws of cutting tools.…”

Improving Tool Wear Prediction with Synthetic Features from Conditional Generative Adversarial Networks

“…Many different DL network architectures have been utilized to predict tool wear from multi-sensor sequences in a supervised learning manner. Such as long-and-short-term memory (LSTM) [10] and its variant gated recurrent units (GRU) [11], convolutional neural networks (CNN) [13], capsule network [14], and Transformer [15]. The strength of CNN for feature extraction in the spatial domain and GRU in the temporal domain was combined in [12].…”