Fault Precognition System for Remaining Useful Life Estimation in Bearing Systems Using Autoencoder-LSTM and Clustering Techniques DOI Creative Commons
Pooja Kamat

Journal Européen des Systèmes Automatisés, Journal Year: 2024, Volume and Issue: 57(6), P. 1721 - 1728

Published: Dec. 31, 2024

This paper proposes a fault precognition system designed for predictive maintenance in bearing systems aimed at improving Remaining Useful Life (RUL) estimation accuracy.This study makes use of the Pronostia-bearing dataset, recognized standard RUL prediction and maintenance.It includes vibration data captured by accelerometer sensors along two axes (X Y), which shows how bearings deteriorate under different operation circumstances.The extensive size several experiencing progressive deterioration, guarantees strong validation suggested actual situations.The utilizes Pronostia employing time-domain feature extraction, automated ranking, pattern classification through Kmeans clustering with Silhouette Coefficients.A core component is an Autoencoder-LSTM model, identifies early occurrences analyzing reconstruction loss thresholds-quantitative measures deviation between observed reconstructed data.These thresholds serve as indicators anomalous behaviour, distinguishing normal operations from fault-prone clusters.The then estimates using various LSTM variants, including Vanilla LSTM, BiLSTM, CNN-LSTM, StackLSTM, ConvLSTM Encoder-Decoder performance evaluated Mean Squared Error (MSE) R² scores.The results demonstrate that incorporating into significantly enhances accuracy, facilitating proactive operational reliability.

Language: Английский

Modeling raster bead deformation process for monitoring fused filament fabrication using acoustic emission DOI Creative Commons
Zhen Li, Lei Fu,

Xinfeng Zou

et al.

Progress in Additive Manufacturing, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 12, 2025

Language: Английский

Citations

0
Development and Experimental Validation of a Hybrid Wire Arc Additive Manufacturing and Milling Repair Platform DOI

S.J. Hu,

Keyi Wang, Xia Li

et al.

International Journal of Precision Engineering and Manufacturing, Journal Year: 2025, Volume and Issue: unknown

Published: March 20, 2025

Language: Английский

Citations

0
LoRe-GRNN: A Hybrid Deep Learning Framework for Real-Time Anomaly Detection and Stress Distribution Prediction in 3D Printing Processes DOI Open Access
Ahmad Joman Alghamdi

Engineering Technology & Applied Science Research, Journal Year: 2025, Volume and Issue: 15(2), P. 21671 - 21677

Published: April 3, 2025

Advanced 3D Printing (A3P) revolutionizes manufacturing with precision, speed, and innovation, unlocking limitless design possibilities superior material performance for next-generation industrial creative applications. A3P epitomizes a paradigm shift in manufacturing, seamlessly merging additive fabrication advanced printing to construct intricate geometries unattainable through conventional methods. However, inherent challenges persist, including structural deformations Stereolithography (SLA) nozzle occlusions Fused Deposition Modeling (FDM), necessitating intelligent intervention. This study introduces LoRe-GRNN, groundbreaking Deep Learning (DL) framework real-time anomaly detection stress distribution prediction. Leveraging novel fusion of Longformer-Reformer (LoRe) architectures Gated Recurrent Neural Networks (GRNN), the system optimizes feature extraction predictive accuracy. A meticulously curated model repository, synergized Finite Element (FE) simulations, enhances SLA predictions, while an integrated multisensory module ensures FDM process monitoring. The hybrid approach demonstrates unparalleled achieving 99.23% accuracy, significantly mitigating computational overhead compared traditional FE simulations. transformative resilience heralding era intelligent, high-fidelity, resource-efficient systems.

Language: Английский

Citations

0
Fault Precognition System for Remaining Useful Life Estimation in Bearing Systems Using Autoencoder-LSTM and Clustering Techniques DOI Creative Commons
Pooja Kamat

Journal Européen des Systèmes Automatisés, Journal Year: 2024, Volume and Issue: 57(6), P. 1721 - 1728

Published: Dec. 31, 2024

This paper proposes a fault precognition system designed for predictive maintenance in bearing systems aimed at improving Remaining Useful Life (RUL) estimation accuracy.This study makes use of the Pronostia-bearing dataset, recognized standard RUL prediction and maintenance.It includes vibration data captured by accelerometer sensors along two axes (X Y), which shows how bearings deteriorate under different operation circumstances.The extensive size several experiencing progressive deterioration, guarantees strong validation suggested actual situations.The utilizes Pronostia employing time-domain feature extraction, automated ranking, pattern classification through Kmeans clustering with Silhouette Coefficients.A core component is an Autoencoder-LSTM model, identifies early occurrences analyzing reconstruction loss thresholds-quantitative measures deviation between observed reconstructed data.These thresholds serve as indicators anomalous behaviour, distinguishing normal operations from fault-prone clusters.The then estimates using various LSTM variants, including Vanilla LSTM, BiLSTM, CNN-LSTM, StackLSTM, ConvLSTM Encoder-Decoder performance evaluated Mean Squared Error (MSE) R² scores.The results demonstrate that incorporating into significantly enhances accuracy, facilitating proactive operational reliability.

Language: Английский

Citations

0