Research on Setting Strategies for Multimodal Faults in Low Voltage Distribution Systems

Le résumé fourni par la source

This study proposes a fixed value setting strategy based on multimodal faults in low-voltage distribution systems. By improving the YOLOv3 algorithm, accurate detection of distribution equipment in infrared images is achieved, and temperature distribution features are extracted. Using the Word2Vec model to process operational text data and construct semantic associated word vectors. Using FFT and STFT transforms to obtain frequency domain features of audio signals. By using feature level fusion strategy to map three modal features to a unified subspace, a multimodal fault analysis model is constructed. Establish setting calculation rules based on expert systems, implement setting value calculation for overcurrent, zero sequence, and other protections in a visual manner, support custom rule adjustment, and automatically generate setting value sheets. The experimental results show that the fault recognition rate of our method in eight sets of tests reached 97.89%–99.67%, and the fixed value error rate remained stable below 1%.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.