Enhanced Fault Diagnosis in Milling Machines Using CWT Image Augmentation and Ant Colony Optimized AlexNet DOI Creative Commons

N. Ullah,

Muhammad Umar,

Jae‐Young Kim

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(23), P. 7466 - 7466

Published: Nov. 22, 2024

A method is proposed for fault classification in milling machines using advanced image processing and machine learning. First, raw data are obtained from real-world industries, representing various types (tool, bearing, gear faults) normal conditions. These converted into two-dimensional continuous wavelet transform (CWT) images superior time-frequency localization. The then augmented to increase dataset diversity techniques such as rotating, scaling, flipping. contrast enhancement filter applied highlight key features, thereby improving the model’s learning detection capability. enhanced fed a modified AlexNet model with three residual blocks efficiently extract both spatial temporal features CWT images. architecture particularly well-suited identifying complex patterns associated different types. deep optimized ant colony optimization reduce dimensionality while preserving relevant information, ensuring effective feature representation. classified support vector machine, effectively distinguishing between conditions high accuracy. provides significant improvements outperforming state-of-the-art methods. It thus promising solution industrial diagnosis has potential broader applications predictive maintenance.

Language: Английский

Feature extraction of combined failures of rolling bearings based on adaptive variance symplectic geometry model decomposition DOI
Mingyue Yu, Ziru Ma, Guanglei Meng

et al.

Journal of Vibration and Control, Journal Year: 2025, Volume and Issue: unknown

Published: March 20, 2025

Vibration signals of rolling bearings with faults are characterized by strong nonlinearity and non-stationarity, making it difficult to extract fault information. In the engineering practice, bearing is often represented as compound. Compared single fault, more feature information combined faults. Symplectic geometric mode decomposition represents better performance can provide protection geometry structure phase space. However, extraction affected ineffective symplectic geometrical components when processing noise weak failure feature. Meanwhile, there a lack effective standard for component option. To solve these problems, an adaptive variance method proposed. decrease interference strengthen features in original signal, sequence signal constructed. prevent influence improper embedding dimension on decomposition, track matrix adaptively determined maximum margin factor criterion. problem being option, optimal activity parameter. Faults identification accomplished power spectrum component. ascertain efficacy superiority proposed method, was compared method. Results indicate that effectively suppress noise, reduce invalid accomplish option components, enables precise judgment bearings. Furthermore, contrast frequencies distributed lower frequency band, which beneficial real-time monitoring applications.

Language: Английский

Citations

0
A power quality disturbance classification method based on improved Shapelet method DOI

Jiabin Luo,

Anqi Jiang,

Shuqing Zhang

et al.

Electric Power Systems Research, Journal Year: 2025, Volume and Issue: 246, P. 111673 - 111673

Published: April 17, 2025

Language: Английский

Citations

0
An optimal filtering frequency band search method based on MZGWO in rolling bearings fault diagnosis DOI
Zejun Zheng, Dongli Song,

Weihua Zhang

et al.

Mechanical Systems and Signal Processing, Journal Year: 2025, Volume and Issue: 232, P. 112773 - 112773

Published: April 25, 2025

Language: Английский

Citations

0
Enhanced Fault Diagnosis in Milling Machines Using CWT Image Augmentation and Ant Colony Optimized AlexNet DOI Creative Commons

N. Ullah,

Muhammad Umar,

Jae‐Young Kim

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(23), P. 7466 - 7466

Published: Nov. 22, 2024

A method is proposed for fault classification in milling machines using advanced image processing and machine learning. First, raw data are obtained from real-world industries, representing various types (tool, bearing, gear faults) normal conditions. These converted into two-dimensional continuous wavelet transform (CWT) images superior time-frequency localization. The then augmented to increase dataset diversity techniques such as rotating, scaling, flipping. contrast enhancement filter applied highlight key features, thereby improving the model’s learning detection capability. enhanced fed a modified AlexNet model with three residual blocks efficiently extract both spatial temporal features CWT images. architecture particularly well-suited identifying complex patterns associated different types. deep optimized ant colony optimization reduce dimensionality while preserving relevant information, ensuring effective feature representation. classified support vector machine, effectively distinguishing between conditions high accuracy. provides significant improvements outperforming state-of-the-art methods. It thus promising solution industrial diagnosis has potential broader applications predictive maintenance.

Language: Английский

Citations

3