Browsing by Author "Zaitsev, Ievgen"

Now showing 1 - 2 of 2

Acoustic-Based Fault Diagnosis of Electric Motors Using Mel Spectrograms and Convolutional Neural Networks
(Nature Portfolio, 2025) Uzel, Hasan; Ozupak, Yildirim; Alpsalaz, Feyyaz; Aslan, Emrah; Zaitsev, Ievgen
This study presents a comprehensive deep learning framework for diagnosing acoustic faults in electric motors. The framework uses Mel spectrograms and a lightweight convolutional neural network (CNN). The method classifies three motor states, engine_good, engine_broken, and engine_heavyload, based on audio recordings from the IDMT-ISA-ELECTRIC-ENGINE dataset. To prevent data leakage and ensure a robust evaluation, the study employed file-level splitting, session separation, 5-fold cross-validation, and repeated trials. The raw audio signals were transformed into Mel spectrograms and processed through a CNN architecture that integrates convolutional, pooling, normalization, and dropout layers. Quantitative metrics, including THD, spectral entropy, and SNR, further characterize the acoustic distinctions between motor states. The proposed model achieved a test accuracy of 99.7%, outperforming or matching state-of-the-art approaches, such as ResNet-18, CRNN, and Transformer classifiers, as well as traditional MFCC-based baselines. Noise robustness and sensitivity analyses demonstrated stable performance under varying SNR conditions and preprocessing settings. Feature-importance maps revealed that low-frequency regions (0-40 Mel bins) were key discriminative components linked to physical fault mechanisms. Computational evaluation confirmed the model's real-time feasibility on embedded hardware with low latency and a modest parameter count. Though primarily validated on one motor type, external-domain testing revealed strong adaptability. Future work may incorporate transfer learning or multimodal fusion. Overall, the proposed framework provides a highly accurate, interpretable, and efficient solution for real-time motor fault diagnosis and predictive maintenance in industrial environments.
Explainable LSTM-AdamW Based Fault Diagnosis of Aircraft Rotating Components Using Airborne Acoustic Signals under Dynamic Operating Conditions
(Nature Portfolio, 2026) Özüpak, Yıldırım; Aslan, Emrah; Zaitsev, Ievgen
Acoustic signal (AS) has emerged as a powerful non-contact technique for early detection of incipient faults in aircraft rotating components due to its high sensitivity to transient damage mechanisms. However, the strong non-stationarity and noise susceptibility of acoustic signals under dynamically varying operating conditions present significant challenges for reliable fault diagnosis. In this study, an explainable deep learning framework based on a Long Short-Term Memory (LSTM) network optimized with the AdamW algorithm is proposed for fault diagnosis of aircraft-related rotating components using acoustic signals. The framework leverages sequential learning to capture the temporal evolution of acoustic signals and is systematically compared with conventional recurrent architectures, including Recurrent Neural Networks (RNNs) and Gated Recurrent Units (GRUs). Experimental results demonstrate that the proposed LSTM-AdamW model achieves superior diagnostic performance, reaching a test accuracy and macro-F1 score of 99.26% under dynamic operating conditions. The enhanced performance is attributed to the LSTM's ability to model long-term temporal dependencies and the regularization benefits of the AdamW optimizer through decoupled weight decay. To improve transparency and physical interpretability, explainable artificial intelligence techniques based on Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) are employed. The explainability analysis reveals that classification decisions are driven by localized, physically meaningful transient acoustic patterns associated with fault-induced events. In addition, a Taylor diagram-based statistical assessment confirms strong agreement between model predictions and reference signals, indicating robust preservation of temporal signal characteristics. The results suggest that the proposed explainable LSTM-AdamW framework provides a reliable, computationally efficient, and interpretable solution for acoustic signal-based fault diagnosis in aerospace applications, with strong potential for real-time condition monitoring and predictive maintenance.