Due to increasing demands for ensuring the safety and reliability of a system, fault detection (FD) has received considerable attention in modern industries to monitor their machines. Bulk materials are transported worldwide using belt conveyors as an essential transport system. The majority of conveyor components are monitored continuously to ensure their reliability, but idlers remain a challenge to monitor due to the large number of idlers (rollers) distributed throughout the working environment. These idlers are prone to external noises or disturbances that cause a failure in the underlying system operations. The research community has begun using machine learning (ML) to detect idler’s defects to assist industries in responding to failures on time. Vibration and acoustic measurements are commonly employed to monitor the condition of idlers. However, there has been no comprehensive review of FD for belt conveyor idlers. This paper presents a recent review of acoustic and vibration signal-based fault detection for belt conveyor idlers using ML models. It also discusses major steps in the approaches, such as data collection, signal processing, feature extraction and selection, and ML model construction. Additionally, the paper provides an overview of the main components of belt conveyor systems, sources of defects in idlers, and a brief introduction to ML models. Finally, it highlights critical open challenges and provides future research directions.

Belt conveyor systems are essential in industries like automotive, aerospace, power generation, and heavy machinery, with idlers playing a crucial role in ensuring the smooth movement of materials. However, constant operation in noisy environments accelerates wear and tear on idlers and obscures early signs of malfunction, such as grinding or rattling from loose parts. This challenge makes early fault detection difficult, increasing downtime and maintenance costs. Therefore, timely and accurate fault detection is vital to prevent severe system issues, ensure optimal performance, and avoid unexpected breakdowns and costly production interruptions. Intelligent Fault Detection (IFD) using artificial intelligence (AI) methods has emerged as a solution, with machine learning techniques like convolutional neural networks (CNNs) proving effective. This study uses YAMNet, initially designed for Sound Event Detection (SED), to identify faults in belt conveyor idlers by analyzing their unique acoustic signatures. We enhance detection capabilities by extracting temporal features from YAMNet-generated embeddings using Bidirectional Long-Term Memory (BiLSTM) and Bidirectional Gated Recurrent Units (BiGRU), augmented with a soft attention mechanism. These features are evaluated using XGBoost, achieving an impressive 90% accuracy in fault detection across idler test sets. Our approach was rigorously compared to the VGGish model and validated on the publicly available MIMII dataset, where it demonstrated superior performance with AUC scores of 0.8355 for fans, 0.9414 for pumps, 0.9265 for valves, and 0.9703 for sliders. These results significantly improve over baseline scores, with increases of 19.36% for fans, 38.44% for pumps, 74.81% for valves, and 38.61% for sliders. This advancement represents a significant step forward in conveyor system diagnostics, providing a robust solution for enhancing industrial safety and operational efficiency.

The shift towards automated inspection methods represents considerable progress in conveyor system maintenance, enhancing safety and efficiency while posing challenges in data analysis and implementation costs. This study critically analyses sensor technologies and inspection methods for detecting faults in conveyor belt idlers, highlighting their essential role in preserving the operational integrity of industrial conveyor systems. By synthesizing various research findings, the study assesses the effectiveness of different sensor devices in identifying defects, including built-in sensors, fixed sensor options like acoustic, ultrasonic sensors, cameras, accelerometers, and Distributed Optical Fibre Sensors (DOFS), as well as mobile sensor systems. Our findings emphasize the accuracy of robot-based systems in identifying bearing defects, the comprehensive coverage provided by drones for medium-scale inspections, the constant monitoring offered by integrated idler sensors, and the ability of fixed sensors to detect mechanical faults despite environmental challenges. This research adds to the ongoing discussion on enhancing conveyor system dependability through technological advancements, providing insights into potential future developments that could further refine maintenance strategies in the sector.

Idlers are essential to conveyor systems, as well as supporting and guiding belts to ensure production efficiency. Proper idler maintenance prevents failures, reduces downtime, cuts costs, and improves reliability. Most studies on idler fault detection rely on supervised methods, which depend on large labelled datasets for training. However, acquiring such labelled data is often challenging in industrial environments due to the rarity of faults and the labour-intensive nature of the labelling process. To address this, we propose the chroma-augmented semi-supervised anomaly detection (CASSAD) method, designed to perform effectively with limited labelled data. At the core of CASSAD is the one-class SVM (OC-SVM), a model specifically developed for anomaly detection in cases where labelled anomalies are scarce. We also compare CASSAD’s performance with other common models like the local outlier factor (LOF) and isolation forest (iForest), evaluating each with the area under the curve (AUC) to assess their ability to distinguish between normal and anomalous data. CASSAD introduces chroma features, such as chroma energy normalised statistics (CENS), the constant-Q transform (CQT), and the chroma short-time Fourier transform (STFT), enhanced through filtering to capture rich harmonic information from idler sounds. To reduce feature complexity, we utilize the mean and standard deviation (std) across chroma features. The dataset is further augmented using additive white Gaussian noise (AWGN). Testing on an industrial dataset of idler sounds, CASSAD achieved an AUC of 96% and an accuracy of 91%, surpassing a baseline autoencoder and other traditional models. These results demonstrate the model’s robustness in detecting anomalies with minimal dependence on labelled data, offering a practical solution for industries with limited labelled datasets.

Fault detection in belt conveyor idlers is crucial for minimising downtime and reducing maintenance costs in industrial operations. Traditional methods, like vibration or temperature-based monitoring, face limitations, including challenging sensor installation and restricted data accessibility. Moreover, these approaches often emphasise spatial features, neglecting the temporal dynamics essential for understanding idler performance over time. This study introduces TD-CLNet, a hybrid fault detection framework that leverages acoustic signals captured via contactless microphones processed through a Time-Distributed CNN-LSTM architecture. The model combines the spatial feature extraction capabilities of Convolutional Neural Networks (CNNs) with the temporal sequence modelling strengths of Long Short-Term Memory (LSTM) networks. A key innovation is the use of the Time-Distributed layer, which enables consistent feature extraction across individual log-Mel spectrogram frames while preserving their temporal relationships. This ensures a robust and coordinated learning process, efficiently addressing the challenges of detecting complementary and relevant features. The performance of TD-CLNet is compared to a frame-based feature extraction approach, which treats each log-Mel spectrogram frame as an independent sample, as well as traditional machine learning methods. Results demonstrate that TD-CLNet achieves a test accuracy of 92% on real-world idler data using K-fold cross-validation, significantly outperforming competing methods. This research provides a scalable and effective solution for fault detection in belt conveyor idlers, advancing predictive maintenance strategies, improving operational efficiency, and minimising unplanned downtime in industrial environments.

Reliable detection of idler faults in belt conveyor systems is essential for minimising unplanned downtime, improving safety, and optimising maintenance. Conventional vibration- and acoustic-based approaches, which use fixed sensors or handheld microphones, often fail under non-stationary, noisy, or spatially variable conditions. To address these challenges, we propose a robust two-phase transfer learning framework leveraging acoustic drone data for cross-domain, non-invasive idler fault diagnosis. The framework combines pretrained Convolutional Neural Network (CNN) backbones, including ResNet-18, DenseNet-121, ShuffleNet-V2, and EfficientNet-B0, with Transformer encoders to capture local spectral and long-range temporal dependencies. Both Mel-spectrograms and MFCCs are employed to extract discriminative spectral-temporal fault features. A Mean-Covariance Alignment (MCA) mechanism is introduced to facilitate domain-invariant feature learning through two sequential phases: (i) source-domain pretraining on handheld recordings to establish robust acoustic representation and (ii) target-domain fine-tuning on drone data using progressive layer unfreezing and discriminative learning rates. Extensive experiments validate the effectiveness of the proposed framework. In Phase one, the ResNet-18–Transformer hybrid achieved a cross-fold mean accuracy on, with a 95% confidence interval ranging from 96.4% to 97.6%. handheld data, Ensemble learning further improved performance to 98.0%, indicating robust feature learning and effective transferability. In phase two, MCA outperformed the cross-entropy (CE) baseline when fully fine-tuning Mel-spectrogram features. The accuracy went from 84.7% (CE) to 86.0%, and the F1-score went from 83.4% to 85.6%. This was better than other domain adaptation methods like MMD, CMD, CORAL, and SWD. While MFCC features achieved higher baseline performance, they generally experienced drops under MCA adaptation. Across adaptation strategies and fine-tuning setups, Mel-spectrogram features demonstrated greater robustness, often maintaining or slightly improving performance, whereas MFCCs were more sensitive to domain adaptation. An ablation study further confirmed that aligning both first- and second-order statistics is crucial for achieving consistent improvements. These results indicate that MCA enhances CNN-Transformer transferability and highlight the reliability of Mel-spectrogram features for handheld-to-drone fault diagnosis in real industrial environments.