Jul 2026· IEEE International Conference on Circuits and Systems for Communications· pp. 1-6· 0 citations· 17 references
Abstract
Brake fault diagnosis is critical for enhancing the safety and reliability of vehicles. This paper presents an approach of utilizing spectrogram deep learning, where the spectrogram is defined as the representation of the signal in terms of timefrequency plane. To diagnose faults in NVH signals, a novel framework combining deep learning models and unsupervised clustering methods is proposed, especially focusing on spectrograms. Raw multi-channel NVH data are divided into overlapping temporal segments and converted into timefrequency spectrogram images via Short-Time Fourier Transform (STFT). Afterward, intrinsic patterns of vibrations are extracted by K-means clustering, and a two-dimensional convolutional neural network (2D CNN) is trained based on the extracted clusters. Experimental datasets are formed by converting NVH signals into 37,962 spectrogram images through 38 sensor channels. Classification accuracy, precision, recall, and F1-score values reached up to 98.41% and more than 0.98 for all clusters, and the areas under ROC curves are close to 1.0 for all clusters.
Rotary machines are vital in industrial and electrical systems, and prompt defect detection is crucial to prevent operational failures and financial losses. This article presents a framework using a Convolutional Neural Network (CNN) for defect detection via spectrogram images derived from simulated voltage, current, and load signals of rotary machines. The dataset, generated using MATLAB simulations and accessible on Kaggle, comprises spectrograms depicting normal operation and three fault conditions: $10 \Omega, 30 \Omega$, and $60 \Omega$. The CNN model proficiently extracts time-frequency characteristics from the spectrograms, attaining an overall classification accuracy of 96.3%, with precision, recall, and F1-scores continuously above 95% across all fault categories. The findings illustrate the model’s capacity to identify nuanced differences in machine behavior resulting from varying fault resistances. In contrast to traditional vibration- and signal-based techniques, the proposed method offers a resilient, non-invasive, and automated alternative for monitoring the state of rotary machines, facilitating predictive maintenance and mitigating the risk of unforeseen breakdowns. This research highlights the efficacy of integrating deep learning with spectrogram analysis for precise industrial problem identification.
R. Vizhi, V. Karthikeyani, L. Sundari et al.· International Conference on...· 0 citations
Electrical current signals provide essential information about the health and performance of electrical systems. Identifying distorted signals is critical for the early detection of faults in electrical systems, and in turn, helps prevent damage, instability, and loss of efficiency. This paper presents a study on classifying healthy and faulty sine-wave signals using convolutional neural networks. A dataset of 200 images was constructed, to provide diverse waveform variations, and used to train and test a customized Convolutional Neural Networks (CNN) in addition to three pretrained CNNs: SqueezeNet, GoogLeNet, and ResNet-50. Each pretrained network was fine-tuned through transfer learning, and data augmentation was applied to improve generalization. Experimental results show that ResNet-50 achieved the highest validation accuracy of 98.33%, while SqueezeNet and GoogLeNet reached 96.67%. Testing on unseen current signal images confirmed that deeper models were more effective in detecting small waveform distortions. The results demonstrate the suitability of CNN-based approaches for waveform classification and highlight the importance of model depth and dataset variation. This study contributes to the field of predictive maintenance by providing an exploration of a simple, cost-effective, and accurate method for fault detection in single phase induction motors. It opens the door for further research into machine learning applications in fault diagnosis in other types of motors and electrical systems.
M. Shatnawi, Mariam Alsaqqaf, Salihah Almenhali et al.· 2026 6th International Confe...· 0 citations
This paper presents a hybrid anomaly detection framework for vibroacoustic signals obtained from a DGEN 380 turbofan test bench. Traditional diagnostic approaches based on statistical thresholding exhibit limited effectiveness in non-stationary environments with multiple interacting vibration and acoustic sources. To address these challenges, a hybrid approach combining unsupervised residual modeling with supervised classification is proposed. The framework is based on the LIRA (Log-amplitude Spectral Residual Autoencoder) model, which learns a baseline representation of the system using only healthy data and identifies deviations through reconstruction-based residual analysis in both time and frequency domains. These residual representations are subsequently transformed into statistical descriptors computed across frequency sub-bands and used as input to a LightGBM classifier. The proposed approach is evaluated on a publicly available vibroacoustic turbofan dataset. The results indicate improved performance compared to classical threshold-based methods, with the boosting model achieving an F1-score of 80.60% and an AUC-ROC of 0.9025. The consistency between cross-validation and test set results further suggests good generalization capability. In addition, the integration of explainable artificial intelligence (XAI) techniques based on SHAP enables interpretation of model decisions. The analysis indicates that energy-related features, particularly RMS, play a dominant role in the detection process, while higher-order statistical descriptors contribute to robustness by capturing non-linear and impulsive behaviors. The proposed framework provides an effective and interpretable solution for anomaly detection in vibroacoustic systems and demonstrates potential for application in predictive maintenance of aeroengines and other rotating machinery.
A Hybrid Knowledge-DL network (HKD-SVM) that utilizes Support Vector Machine (SVM) as the classifier, making the network well-suited for nonlinear, high-dimensional classification scenarios with limited training samples, which are common in power system applications.
Reliable bearing fault diagnosis is essential for predictive maintenance of rotating machinery, particularly in applications where contact-based vibration sensors are difficult to install or maintain. This study proposes an explainable acoustic fault diagnosis framework based on Mel-spectrogram representations and a hybrid Convolutional Neural Network–Gated Recurrent Unit (CNN–GRU) architecture. Acoustic emission signals collected from a controlled bearing fault simulator are first segmented and transformed into Mel-spectrograms to represent the time–frequency structure of normal and faulty bearing conditions. The CNN module extracts localized spectral patterns from the Mel-spectrogram images, while the GRU module models temporal dependencies along the spectrogram time axis with fewer recurrent parameters than conventional LSTM-based designs. To address the interpretability limitations of deep learning models, Grad-CAM, Integrated Gradients, and SHAP are employed to analyze the frequency–time regions contributing to the model decisions. In addition, the explanation maps are compared with fault-relevant spectral regions derived from fault characteristic frequency analysis to evaluate whether the model focuses on physically meaningful patterns rather than arbitrary image regions. The proposed framework achieved an high accuracy of in the initial experiment and was further evaluated through repeated validation to assess performance stability under small-sample conditions. The results demonstrate that acoustic sensing combined with explainable CNN–GRU learning can provide a non-contact and interpretable alternative for bearing fault diagnosis. However, the limited dataset size remains an important constraint, and future studies should validate the framework on larger record-level and cross-domain datasets.
Emrah Aslan, Yıldırım Özüpak· Proceedings of the Instituti...· 1 citation
The availability and reliability of the power transmission system are crucial to ensuring the continuity of the electrical energy supply. Disturbances caused by natural events, vegetation interference, or animal activity can lead to widespread blackouts, necessitating rapid and accurate fault analysis. This study proposes a multi-representation deep learning framework for fault classification using a hybrid Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture driven by raw Disturbance Fault Recorder (DFR) data. To ensure rigorous evaluation and strictly prevent data leakage, an initial dataset of 457 raw DFR recordings is partitioned using a stratified split method. Exactly 87 validation and 54 testing samples are strictly isolated to preserve their real-world integrity, while the remaining 316 training samples are synthetically augmented to a perfectly balanced 1,500 samples (500 per class). These are processed seamlessly into 1D time-series signals and 2D stacked spatial images at a 224x224 resolution. Furthermore, an Uncertainty Rejection strategy utilizing a 60% confidence threshold is implemented to dynamically intercept ambiguous transient anomalies and prevent forced misclassifications. Experimental results demonstrate that the proposed hybrid CNN-LSTM model achieves an overall classification accuracy of 96.3%. While this 3.7% accuracy improvement corresponds to exactly a 2-sample difference on the constrained testing set compared to the established standalone baselines (LSTM and CNN, both at 92.6%), it serves as preliminary evidence demonstrating that the hybrid architecture can resolve specific, ambiguous edge-cases that single-view models fail to classify. The proposed framework offers a reliable, visually-explainable foundation that can be implemented in a monitoring system to assist operators in making faster and more accurate decisions in fault handling.
Hafizh Tri Januar, D. Purwanto, D. Kuswidiastuti· International Seminar on Int...· 0 citations