基于改进SVM的电网设备状态预测与评估模型设计  

Design of Power Grid Equipment Condition Prediction and Evaluation Model Based on Improved SVM

作　　者：孙文成 周林 任成君 李卓雯 彭宇辉 SUN Wencheng;ZHOU Lin;REN Chengjun;LI Zhuowen;PENG Yuhui(Southwest Branch of State Grid Corporation of China,Chengdu 610000,China)

机构地区：[1]国家电网有限公司西南分部,四川成都610000

出　　处：《微型电脑应用》2025年第2期47-50,共4页Microcomputer Applications

基　　金：西南分部基金(C00J-300009278-00015)。

摘　　要：随着信息技术研究的发展,智能化电网设备所采集的数据得到了有效的储存与利用,为电网设备状态的预测和评估提供技术支撑。在这个背景下,开展了基于改进SVM(support vector machihe)的电网设备状态预测与评估模型设计研究。研究结合CNN模型,在SVM算法的基础上通过转换思维将模型的求解变更为线性问题,并设置了核函数以降低样本数据的维数。在处理异常冗余数据时,则采用极值分析法对处于拐点的数据进行处理,并在人工神经网络迭代时删除凸包拐点数据。同时,基于改进SVM算法选择状态量来建立设备的相关体系,进而构建设备状态的评估模型。算例分析结果表明,在不同类型数据集上所设计模型的预测精度均在95%以上,召回率及F 1值也达到了预期目标。With the development of information technology research,the data collected by intelligent power grid equipment can be effectively stored and utilized,and provide technical support for power grid equipment status prediction and evaluation.For this reason,this paper carries out the research on the design of power grid equipment condition prediction and evaluation model based on improved SVM(support vector machihe).Combined with the CNN model,based on the SVM algorithm,the problem is transformed into a linear problem through transformation thinking,and the kernel function is set to reduce the dimension of the sample data.When dealing with abnormal redundant data,the extreme value analysis method is used to deal with the data at the inflection point,and the convex hull inflection point data are deleted during the artificial neural network iteration.Based on the improved SVM algorithm,the equipment status quantity system is established and the equipment status evaluation model is constructed.The results of the example analysis show that the prediction accuracy of the designed model is above 95%under different types of data sets,and the recall rate and F 1 value reach the expected goal.

关 键 词：改进SVM 深度学习 状态预测 数据处理 

分 类 号：TP807[自动化与计算机技术—检测技术与自动化装置] TN929.5[自动化与计算机技术—控制科学与工程]