基于改进型神经网络的变电站监控信息点表生成优化研究  

Research on the Transformer Substation Monitoring Information TableGeneration Optimum Based on Modified Neural Network

作　　者：刘伟 王迟 刘晓天 左晨宇 李阔 LIU Wei;WANG Chi;LIU Xiao-tian;ZUO Chen-yu;LI Kuo(State Grid Daxing Power Supply Company of Beijing,Beijing 100026,China)

机构地区：[1]国网北京大兴供电公司,北京100026

出　　处：《电气开关》2025年第2期35-37,共3页Electric Switchgear

摘　　要：由于现有的优化方法信息量增益结果较小,为此研究基于改进型神经网络的变电站监控信息点表生成优化。通过对监控信息点表建立监控信息点表之间的映射关系。计算网络输出与期望目标之间的均方误差来量化全局误差,以此衡量网络的学习效果。通过改进算法进行数据压缩,获取BNN信息传递的最优输出期望值,以此来达到优化的目的。实验结果表明,三个小组的信息量增益结果均在130以上,结果符合预期要求。说明运用此优化监控信息电表的生成过程,可以减少冗余信息,使监控数据更加有效。Study transformer substation monitering information table generation optimum based on modified neural network.By setting up mapping relation between the monitoring information tables,calculate mean-square error between network output and expected target to quantize global errors,For this reason,measure the network′s study effect,By modified algorithm,carry out data compression,achieve optimal output expected value of BNN information transmission,for this reason,achieve optimization purpose.The experimental result shows that three graps of information content gain results are above 130.The result is up to expected requirements,making monitoring data more effective.

关 键 词：神经网络 改进 变电站 监控信息 生成点表 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]