摘要
传统识别方法在分类过程中提取数据特征存在次序性,导致台区户变关系识别结果误差较大,为此,提出了基于BiLSTM-TCNN的台区户变关系自动识别方法。采用远程集中抄表的方式获取台区用户的用电数据,并对原始数据进行预处理后作为样本数据,构建一种由BiLSTM和TCNN组成的并行结构分类模型,将样本数据输入模型,经训练后输出台区户变关系自动识别结果。结果表明,该设计方法识别台区户变关系时皮尔逊相关系数值为0.98,证实了该方法具有较高的识别精度。
Traditional recognition methods extract data feature s in a sequential manner during the classification process,which leads to significant errors in the recognition results of substation household relationships.Therefore,a BiLSTM-TCNN based automatic recognition method for substation household relationships is proposed.Using remote centralized meter reading to obtain electricity consumption data of substation users,preprocessing the original data as sample data,constructing a parallel structured classification model composed of BiLSTM and TCNN,inputting the sample data into the model,and outputting the automatic recognition results of substation user relationship after training.The results indicate that the Pearson correlation coefficient value for identifying the household variation relationship in the substation area using this design method is 0.98,confirming that the method has high recognition accuracy.
作者
黄继盛
李光明
周善善
谢宗禄
HUANG Jisheng;LI Guangming;ZHOU Shanshan;XIE Zonglu(Lincang Power Supply Bureau of Yunnan Power Grid Co.,Ltd.,Lijiang,Yunnan 677000,China)
出处
《自动化应用》
2024年第7期230-232,共3页
Automation Application
基金
中国南方电网有限责任公司科技项目(YNKJXM20222483)。