Skip to content
Open access

A Hybrid CNN-BiLSTM-Multihead Attention Framework for Robust Broken Tooth Fault Diagnosis in Gearboxes Under Noise Conditions

Oct 2026 · U Porto Journal of Engineering · Vol 12, pp. 132-155 · 0 citations · 13 references

TL;DR

A hybrid deep learning model combining 1D CNN, Bidirectional LSTM, and Multi-Head Attention mechanism to classify healthy and fractured tooth conditions using vibration signals obtained from the SpectraQuest simulator demonstrates that the model avoids overlearning and adapts to varying operating conditions in industrial environments.

Abstract

Early detection of defects in rotary machinery is critical to preventing system downtime and cost losses. This study proposes a hybrid deep learning model combining 1D CNN, Bidirectional LSTM, and Multi-Head Attention mechanism to classify healthy and fractured tooth conditions using vibration signals obtained from the SpectraQuest simulator. The model, which works directly on raw signals, eliminates manual feature extraction while achieving noise resilience through rich data enhancement techniques such as amplitude scaling and frequency masking. Training results show that training, validation and testing accuracy reached 99.92%, 97.00% and 97.13%, respectively. This approximately 2.8% generalization difference demonstrates that the model avoids overlearning and adapts to varying operating conditions in industrial environments. These achievements have been supported by 5-fold cross-validation and are presented in detail and comparatively through durability tests performed at different noise levels (5-20 dB SNR) and ablation studies revealing the contribution of architectural components.

Read PDF

Similar papers

Conference Oct 2026

TinyCaracalFormer: a lightweight frequency-domain hybrid model for steel surface defect inspection

Manual inspection in steel production suffers from low efficiency and unstable accuracy, while standard Transformer models are limited by quadratic self-attention complexity and mandatory positional encoding. This work constructs a unified two-stage steel defect detection framework with dual convolutional neural networ...

Yao-Hui Shen, Cai-Yun Chen · 0 citations
Oct 2026

A novel bearing fault diagnosis approach combining F-VMD with CNN-BiLSTM and cross-attention

To enhance the reliability and accuracy of bearing fault diagnosis under challenging operating conditions, this study proposes a method that integrates a combined fast Fourier transform (FFT) and variational mode decomposition (VMD) preprocessing framework with a convolutional neural network (CNN)-bidirectional long sh...

Jie Li, Yu-Ying Wang, Jie Yang et al. · 0 citations
Open access Sep 2026

Deep Learning-Based Visual Detection Technology for Metal Gear Surface Defects

Due to complex background interference, a lack of real defect samples, and inconsistent evaluation of real-time deployment in metal gear surface defect detection, four representative studies were selected for comparison and four technical approaches were analyzed: lightweight detection, interference resistance, few-sam...

Jie Meng · 0 citations
Conference Sep 2026

Gearbox multimodal fault diagnosis based on microphone array and cross-attention

In order to solve the problems of limited vibration signal acquisition, incomplete single-modal features, and sound signals being vulnerable to noise interference in gearbox fault diagnosis, a Multi-Channel Convolutional-Transformer Cross- Attention Diagnosis Model is proposed. Its core innovations are as follows: comb...

Jia-Shun Deng, Li-Juan Ji, Yu Gong et al. · 1 citation
Conference Sep 2026

Gearbox fault diagnosis and feature visualization based on multiscale convolution network with LSTM and attention mechanism

An end-to-end fault diagnosis model combining Multi-Scale Convolutional Neural Network, MSCNN, Long Short-Term Memory (LSTM) network and attention mechanism is proposed to improve diagnostic accuracy and model interpretability for gearbox fault diagnosis.

Tong Fang, Xiang Sun · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.