Inspection of magnetic tile internal cracks based on impact acoustics

Abstract

An automatic system is developed for internal cracks detection in magnetic tiles based on the impact acoustics, using wavelet packet transform (WPT), principal component analysis (PCA) and hidden Markov model (HMM). In this system, the detecting device is considered as core part to collect and analyse the impact sounds. The original impact sounds are first decomposed up to six levels based on WPT to extract the features. PCA is then performed for dimension reduction and clustering analysis. By adopting the features extracted based on WPT and optimised by PCA as inputs, an HHM classifier is developed for automatic inspection. The results of classification show that the accuracy rate is 100%, demonstrating that the system has significant potential in detecting magnetic tile internal cracks.

Real-time hollow defect detection in tiles using on-device tiny machine learning

“…Our investigation substantially enhances existing methods in tile defect detection, as evidenced by a comprehensive comparison with prior studies [5][6][7][8][9][10][11][12][13][14] delineated in table 3. Our approach introduces a holistic improvement in several key performance indicators, warranting a detailed discussion on each advancement.…”

“…Accurate and realistic performance metrics: While studies [7,8] report 100% accuracy in laboratory conditions, our research shows a competitive accuracy of 97% for validation, 92.48% for testing, and a real-world measured accuracy of 81.25%. These figures represent a realistic performance in practical scenarios, a disclosure often omitted in academic literature.…”

“…Acknowledging the limitations of current methodologies, such as the need for multiple devices and the high costs associated with AI-based diagnostics [7,8,13], the research community has explored the potential of tiny machine learning (TinyML), a technology that combines machine learning with microcontroller units (MCUs) [33][34][35][36][37]. TinyML optimizes models for devices with limited computational resources, enabling the deployment of machine-learning models on a smaller scale.…”

Combined Convolutional and LSTM Recurrent Neural Networks for Internal Defect Detection of Arc Magnets Under Strong Noises and Variable Object Types

“…To verify the advantages of the proposed WConv-LSTM arc magnets internal defective detection model under complex interference conditions compared with traditional machine learning and common end-to-end fault diagnosis methods, we conducted experiments using WPT, HDM, PCA, VMD, WConv-Attention, GRU, Transformer, and WDCNN. Among them, WPT [1], HDM [2], PCA [3], VMD [4] are all traditional machine learning methods for the internal defect detection of arc magnets proposed by our team. WConv-Attention is a hybrid network model that uses the attention mechanism to replace the LSTM of our model.…”

“…The nonlinearity and instability of acoustic signals can seriously hinder the extraction and identification of the signal features regarding internal defects, whereas these meaningful features are usually too weak to be discovered. For this reason, Qinyuan Huang, Luofeng Xie, Yue Zhao, Ming Yin et al successively used wavelet packet analysis (WPT) [1], hidden Markov model (HDM) [2], principal component analysis (PCA) [3], variational mode decomposition (VMD) [4], and other technologies process the acoustic signals of the arc magnet and have achieved a better diagnostic result. However, using the above methods to extract features and diagnose defects within arc magnets generally requires complex mathematical operations and a certain understanding of the extracted signals as well as a wealth of signal processing knowledge.…”