基于RELS-TSVM的电能表电压异常数据识别

为提高电能计量工作中高压用户电能表电压异常数据识别的快速性和准确性,以负荷控制模块深度分析功能所筛选出的电压数据作为研究对象,结合鲁棒能量模型-最小二乘双支持向量机分类方法,采用蝙蝠算法对分类器参数进行优化,实现对电压异常数据的自动判断。实验结果表明,本文所提出的分类模型在泛化性能和鲁棒性方面具有较大改进,在10折交叉验证过程中对于100V、57.7V、220V额定电压等级下的电压异常数据平均识别准确率均达到100%,相比于传统的经验公式判别、决策树分类模型以及简单支持向量机分类方法在分类准确率上分别提升12.53%、3.8%、1.73%。验证了基于鲁棒能量模型-最小二乘双支持向量机分类方法的电压异常数据识别方案的可行性和优势,为基于大数据分析的计量在线监测相关研究提供了新的思路。

Recognition of power meter voltage anomaly data based on RELS-TSVM

In order to improve the speed and accuracy of identifying the abnormal voltage data of high-voltage power meters in power metering,the load control module is used as the depth of the power meter.The voltage data filtered by the analysis function is used as the object of study in combination with the robust energy model-least-squares double support vector machine classification method.The bat algorithm is used to optimize the classifier parameters to achieve automatic determination of the voltage anomaly data.The experimental results show that the classification model proposed in this paper has a significant improvement in generalization performance and robustness,and it is validated in a 10-fold crossover.In the process for 100V,57.7V,220V rated voltage level of voltage anomaly data average identification accuracy rate are up to 100%,compared to traditional empirical formula discrimination,decision tree classification model and simple support vector machine classification methods in classification.The accuracy is improved by 12.53%,3.8%and 1.73%,respectively.The feasibility of the voltage anomaly data identification scheme based on the robust energy model-least-squares double support vector machine classification method is verified and the advantages provide new ideas for research related to online monitoring of metrology based on big data analysis.