Algorithm optimization based on intelligent management of computer electrical equipment: A comprehensive method for PT power monitoring and remote fault indicator

• Published in: Journal of computational methods in sciences and engineering Vol. 25; no. 3; pp. 2577 - 2590
• Main Authors: Wang, Peng, Liu, Wang, Sun, Jun, Chen, Liang, Lu, Chen, Zheng, Bowen
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01.05.2025; Sage Publications Ltd
• Subjects: Accuracy; Algorithms; Back propagation; Electric equipment; Electric power systems; Entropy (Information theory); Fast Fourier transformations; Fault detection; Fault diagnosis; Fourier transforms; Frequency drift; Knowledge bases (artificial intelligence); Multilayers; Optimization; Pattern matching; Real time; Remote monitoring; Smart grid; Voltage drop; Wavelet transforms; remote fault indicator; electromechanical equipment management; power monitoring; fault detection; long short-term memory
• ISSN: 1472-7978; 1875-8983
• DOI: 10.1177/14727978251314564

In response to the accuracy limitations and response delays in current power and fault detection of electromechanical equipment, long short-term memory (LSTM) network algorithm was applied for intelligent optimization and management. Firstly, voltage and power data were collected through a potential transformer (PT), and wavelet transform (WT) was applied to remove noise. Fast Fourier transform (FFT) was utilized to extract key features. Secondly, a multi-layer long short-term memory network was designed, and back propagation algorithms and time series power data were used for LSTM network training to analyze abnormal fluctuations and trends in power data in real-time, and adjust threshold settings. Then, combining the model output and historical fault data, a fault mode knowledge base was established. Potential faults were determined through pattern matching, and signals were sent out through remote indicators. Finally, the algorithm model was evaluated. The research results showed that the weighted response time of voltage drop faults was shortened by 3.3%, and the information entropy values of nine experiments were distributed from 5.11 to 5.28. The diagnostic accuracy for frequency drift was improved by 11.1%. The comprehensive algorithm model used can effectively improve the accuracy of power monitoring and response speed. This study optimizes power monitoring and fault diagnosis by applying the LSTM network algorithm, addressing the limitations of existing methods in terms of real-time performance and accuracy. It effectively enhances the fault prediction capability and response speed of power systems, offering significant application value in the fields of smart grids and equipment management.