A Brief Review of Acoustic and Vibration Signal-Based Fault Detection for Belt Conveyor Idlers Using Machine Learning Models

Sensors, Год журнала: 2023, Номер 23(4), С. 1902 - 1902

Опубликована: Фев. 8, 2023

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

Язык: Английский

---

Data Augmentation and Deep Learning Methods in Sound Classification: A Systematic Review

Electronics, Год журнала: 2022, Номер 11(22), С. 3795 - 3795

Опубликована: Ноя. 18, 2022

The aim of this systematic literature review (SLR) is to identify and critically evaluate current research advancements with respect small data the use augmentation methods increase amount available for deep learning classifiers sound (including voice, speech, related audio signals) classification. Methodology: This SLR was carried out based on standard guidelines PRISMA, three bibliographic databases were examined, namely, Web Science, SCOPUS, IEEE Xplore. Findings. initial search findings using variety keyword combinations in last five years (2017–2021) resulted a total 131 papers. To select relevant articles that are within scope study, we adopted some screening exclusion criteria snowballing (forward backward snowballing) which 56 selected articles. Originality: Shortcomings previous studies include lack sufficient data, weakly labelled unbalanced datasets, noisy poor representations features, effective approach affecting overall performance classifiers, discuss article. Following analysis identified articles, overview feature extraction methods, techniques, its applications different areas classification problem. Finally, conclude summary SLR, answers questions, recommendations task.

Язык: Английский

---

Prediction of Machine Failure in Industry 4.0: A Hybrid CNN-LSTM Framework

Applied Sciences, Год журнала: 2022, Номер 12(9), С. 4221 - 4221

Опубликована: Апрель 22, 2022

The proliferation of sensing technologies such as sensors has resulted in vast amounts time-series data being produced by machines industrial plants and factories. There is much information available that can be used to predict machine breakdown degradation a given factory. downtime equipment accounts for heavy losses revenue reduced making accurate failure predictions using the sensor data. Internet Things (IoT) have made it possible collect real time. We found hybrid modelling result efficient they are capable capturing abstract features which facilitate better predictions. In addition, developing effective optimization strategy difficult because complex nature different time scenarios. This work proposes method multivariate forecasting predictive maintenance (PdM) based on combination convolutional neural networks long short term memory with skip connection (CNN-LSTM). experiment CNN, LSTM, CNN-LSTM models one prediction failures. this from Microsoft’s case study. dataset provides about history, error conditions, telemetry, consists voltage, pressure, vibration, rotation values recorded between 2015 2016. proposed framework two-stage end-to-end model LSTM leveraged analyze relationships among variables through its function, 1-D CNNs responsible extraction high-level Our learns long-term patterns series extracting short-term dependency variables. our evaluation, provided most reliable highest accuracy.

Язык: Английский

---

State-of-the-Art Review of Machine Learning Applications in Additive Manufacturing; from Design to Manufacturing and Property Control

Archives of Computational Methods in Engineering, Год журнала: 2022, Номер 29(7), С. 5663 - 5721

Опубликована: Июль 22, 2022

Язык: Английский

---

Mel Spectrogram-based advanced deep temporal clustering model with unsupervised data for fault diagnosis

Expert Systems with Applications, Год журнала: 2023, Номер 217, С. 119551 - 119551

Опубликована: Янв. 13, 2023

Fault diagnosis of mechanical equipment using data-driven machine learning methods has been developed recently as a promising technique for improving the reliability industrial systems. However, these suffer from data sparsity due to difficulty in collection, which limits feature extraction anomalies. To solve this problem, we propose mel spectrogram-based advanced deep temporal clustering (ADTC) model, can extract and verify features unlabeled through an unsupervised based autoencoder K-means. In addition, ADTC model uses proposed centroid obtain calibrated by minimizing point target distances misclustered encoder output ensemble-based learning. The classifier supervised support vector network is robust nonlinear data, diagnose faults equipment. was validated dataset with augmentation address imbalanced problem. During experiments, exhibited best performance various environment prediction accuracy high 98.06%, outperforming other compared algorithms.

Язык: Английский

---

A novel application of deep transfer learning with audio pre-trained models in pump audio fault detection

Computers in Industry, Год журнала: 2023, Номер 147, С. 103872 - 103872

Опубликована: Фев. 7, 2023

Язык: Английский

---

Robotics Perception and Control: Key Technologies and Applications

Micromachines, Год журнала: 2024, Номер 15(4), С. 531 - 531

Опубликована: Апрель 15, 2024

The integration of advanced sensor technologies has significantly propelled the dynamic development robotics, thus inaugurating a new era in automation and artificial intelligence. Given rapid advancements robotics technology, its core area—robot control technology—has attracted increasing attention. Notably, sensors fusion technologies, which are considered essential for enhancing robot have been widely successfully applied field robotics. Therefore, techniques with enables adaptation to various tasks situations, is emerging as promising approach. This review seeks delineate how combined technologies. It presents nine types used control, discusses representative methods, summarizes their applications across domains. Finally, this survey existing challenges potential future directions.

Язык: Английский

---

Priori-distribution-guided adaptive sparse attention for cross-domain feature mining in diesel engine fault diagnosis

Engineering Applications of Artificial Intelligence, Год журнала: 2024, Номер 132, С. 107975 - 107975

Опубликована: Фев. 1, 2024

Язык: Английский

---

ScaloAdaptAlert, a novel framework for supervised anomaly detection in industrial acoustic data, integrating power scalograms, adaptive filter banks, and convolutional neural networks — A case study

Journal of Manufacturing Systems, Год журнала: 2025, Номер 79, С. 234 - 254

Опубликована: Янв. 31, 2025

Язык: Английский

---

The Method of Rolling Bearing Fault Diagnosis Based on Multi-Domain Supervised Learning of Convolution Neural Network

Energies, Год журнала: 2022, Номер 15(13), С. 4614 - 4614

Опубликована: Июнь 23, 2022

The rolling bearing is a critical part of rotating machinery and its condition determines the performance industrial equipment; it necessary to detect faults as early possible. traditional methods fault diagnosis are not efficient time-consuming. With help deep learning, convolution neural network (CNN) plays huge role in data-driven diagnosis. However, vibration signal non-stationary, contains high noise, one-dimensional, which difficult analyze directly by CNN model. Considering multi-domain learning an advantage this paper proposes novel approach using improved one-dimensional (1D) two-dimensional (2D) two-domain information learning. constructed model combining 1D 2D extracts features from samples. padding dropout technology utilized fully extract raw data reduce over-fitting. To prove validity proposed method, performs two tests with datasets, Case Western Reserve University (CWRU) dataset Dalian Technology (DUT) laboratory dataset. experimental results show that our method achieves recognition accuracy states via monitoring data, there no manual experience necessary. Vibration under strong noise were also used test superiority robustness method.

Язык: Английский

---