基于卷积神经网络的线路损耗智能评估方法研究  

Research on Intelligent Evaluation of Line Loss Based on Convolutional Neural Network

作　　者：肖荣洋 黄雁 XIAO Rongyang;HUANG Yan(Longyan Power Supply Company,State Grid Fujian Electric Power Co.,Ltd.,Longyan,364000,China)

机构地区：[1]国网福建省电力有限公司龙岩供电公司,福建龙岩市364000

出　　处：《自动化与仪器仪表》2023年第1期188-193,共6页Automation & Instrumentation

基　　金：国网福建省电力有限公司科技项目《应对‘双碳’挑战下新型电网的电能质量损耗量化评估及降损减排方案研究与应用》(521360220001)。

摘　　要：针对传统线路损耗评估准确率,导致电缆线路损耗增加的问题,基于卷积神经网络CNN,提出一种关于电缆线路损耗的智能评估方法,并计及电能质量问题,构建一个智能评估模型CPQ-CNN。该模型以电缆线路的电压、电缆和电能质量指标作为CNN的输入向量;进行CNN特征提取和处理后,最终输出电缆线路损耗评估值。仿真结果表明,相较于Un-CPQ-CNN模型,本模型的RMSE和MAE分别降低了22.58%和34.51%,对比于ELM模型、SVR模型和赋权法模型,本模型的RMSE和MAE分别下降了39.05%、40.84%、77.48%和49.42%、44.52%、81.51%。综合分析可知,本模型对电缆线路损耗的智能评估值与真实值间的误差最小,评估准确率更高,可在电缆线路损耗评估中进行应用。In view of the problem of the traditional line loss evaluation accuracy and the increase of cable line loss,an intelligent evaluation method of cable line loss is proposed based on the convolutional neural network CNN,and then the power quality problem is calculated to build an intelligent evaluation model CPQ-CNN.The model takes the voltage,cable and power quality index of the cable line as the input vector of the CNN;after performing the CNN features,the final output cable line loss evaluation value is extracted and processed.The simulation results showed that the RMSE and MAE decreased by 22.58%and 34.51%,respectively,compared to the Un-CPQ-CNN model,and the RMSE and MAE decreased by 39.05%,40.84%,47.48%and 49.52%,44.52%,and 81.51%,respectively.The comprehensive analysis shows that this model has the smallest error between the intelligent evaluation value and the real value,which can be applied in the cable line loss evaluation.

关 键 词：卷积神经网络 电缆线路损耗 智能评估 电能质量指标 质量扰动 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]