基于神经网络算法的10kV配网差异化运维策略研究

Research on Differential Operation and Maintenance Strategy of 10kV Distribution Network Based on Neural Network Algorithm

传统基于主观性权重的差异化运维方法一般运用在高压输电线路中,而广州电网配网线路数量及总长度是其输电线路的10倍以上,加之配网线路建设基础薄弱、沿线情况复杂,相对输电线路对运维人力资源需求更加迫切,因此提出了一种基于神经网络的AI算法。根据考虑权重因子建立的线路分级模型,利用神经网络算法收敛性好、主动学习的特性,在对典型配网线路进行学习后运用于其他线路上,并将结果成功应用于某区配网线路实际运维中,在运维人力资源有限的约束下,优化了每天的运维工作量,极大提高了运维效率,提升了线路整体健康程度,减少了线路故障跳闸率。

The traditional differential operation and maintenance method based on subjective weight is generally used in high voltage transmission lines.The number and total length of distribution lines in Guangzhou power network is more than 10 times that of its transmission lines,in addition,the construction of distribution network line is weak and the situation along the line is complex,so the demand of human resources for operation and maintenance is more urgent.In this paper,a neural network-based Ai Algorithm is proposed based on the above-mentioned background.According to the line classification model considering the weight factor,the neural network algorithm has good convergence and active learning characteristics.After learning the typical distribution network lines,it is applied to other lines,and the results are successfully applied to the actual operation and maintenance of a distribution network line.Under the constraint of limited human resources,the daily operation and maintenance workload is optimized,the operation and maintenance efficiency is greatly improved,the overall health of the line is improved,and the line fault trip rate is reduced.