摘要
配电系统易受外部环境影响,在时空尺度上对潜在故障风险进行分类,对于制订配电系统维护与检修计划,以提高整个配电系统的可靠性至关重要。针对配电系统数据的时空分布不平衡问题,提出了一种基于Transformer模型和注意力机制的配电系统故障风险动态分类模型,旨在对故障结果、故障原因和故障位置进行预测,并结合多种特征属性进行综合分析。实验结果表明,该模型在不同故障的分类任务中表现优异,特别是在捕捉自然因素引起的故障方面展现出显著优势。这项研究不仅进一步证实了自注意力机制和位置编码在提高模型性能方面的核心作用,还充分体现了它们在配电系统数据分析领域中的实用价值。
The distribution system is susceptible to external environmental influences.Classifying potential failure risks in time and space is crucial for developing maintenance and overhaul plans for the distribution system to improve the reliability of the entire distribution system.A dynamic classification model for distribution system fault risk based on Transformer model and attention mechanism is proposed to address the problem of imbalanced spatiotemporal distribution of distribution system data.The aim is to predict the fault results,causes,and locations,and conduct a comprehensive analysis based on multiple feature attributes.The experimental results show that the model performs well in different fault classification tasks,especially in capturing the faults caused by natural factors,showing the significant advantages.This study not only further confirms the crucial role of self-attention mechanism and position encoding in improving model performance,but also fully demonstrates their practical value in data analysis of distribution systems.
作者
安业腾
史嘉琪
谢青
AN Yeteng;SHI Jiaqi;XIE Qing(Customer Service Center of State Grid Corporation,Tianjin 300306,China;Beijing Zhongdian Puhua Information Technology Company Limited,Beijing 100031,China)
出处
《仪表技术》
2024年第5期77-80,共4页
Instrumentation Technology
