Enhanced Fault Diagnosis in Motor Bearings: Leveraging Optimized Wavelet Transform and Non-Local Attention
Journal Title: International Journal of Knowledge and Innovation Studies - Year 2023, Vol 1, Issue 2
Abstract
Recent advancements in non-destructive testing methodologies have significantly propelled the efficiency of bearing defect detection, vital for maintaining optimal final quality standards. This study introduces a novel approach, integrating an Optimized Continuous Wavelet Transform (OCWT) and a Non-Local Convolutional Block Attention Module (NCBAM), to elevate fault diagnosis in motor bearings. The OCWT, central to this methodology, undergoes fine-tuning through a newly formulated metaheuristic algorithm, the Skill Optimization Algorithm (SOA). This algorithm bifurcates into two critical components: the acquisition of expertise (exploration) and the enhancement of individual capabilities (exploitation). The NCBAM, proposed for classification, adeptly captures long-range dependencies across spatial and channel dimensions. Furthermore, the model employs a learning matrix, adept at synthesizing spatial, channel, and temporal data, thus effectively balancing diverse data contributions by extracting intricate interrelations. The model's efficacy is rigorously validated using a gearbox dataset and a motor bearing dataset. The outcomes reveal superior performance, with the model achieving an average accuracy of 94.17% on the bearing dataset and 95.77% on the gearbox dataset. These results demonstrably surpass those of existing alternatives, underscoring the model's potential in enhancing fault diagnosis accuracy in motor bearings.
Authors and Affiliations
Syed Ziaur Rahman, Ramesh Vatambeti
Keywords
Related Articles
- EP ID EP732608
- DOI https://doi.org/10.56578/ijkis010205
- Views 82
- Downloads 0
Ontology-Based Method for Constructing Process Knowledge Models of Aircraft Engine Components
This study addresses the issues of fragmentation, unstructured information, and low reusability in the process knowledge management of aircraft engine component manufacturing. A process knowledge modeling method based on...
AMBERT-DWPM: An Adaptive Masking and Dynamic Prototype Learning Framework for Few-Shot Text Classification
Transformer-based language models have demonstrated remarkable success in few-shot text classification; however, their effectiveness is often constrained by challenges such as high intraclass diversity and interclass sim...
Racism and Hate Speech Detection on Twitter: A QAHA-Based Hybrid Deep Learning Approach Using LSTM-CNN
Twitter, a predominant platform for instantaneous communication and idea dissemination, is often exploited by cybercriminals for victim harassment through sexism, racism, hate speech, and trolling using pseudony-mous acc...
Selection of Enhanced Security Systems Using Complex T-Spherical Fuzzy Models Within a Complex Fuzzy Environment
The theory of Complex T-Spherical Fuzzy Sets (CTSpFSs) is introduced along with their Einstein operational methods under induced variables. This research aims to extend the theoretical framework of complex fuzzy sets (CF...
Selection of CRM Systems Using Objective Criteria for the Needs of Small Companies
This research examines customer relationship management (CRM) systems using multi-criteria decision-making (MCDM) methods, with the intention of selecting the most suitable solution for small companies. The main goal of...