摘要
为了提升家庭智能用电管理系统的计算效能,使用时序分析算法,将远程抄表实时费控电能表采集的1s步长的电流、电压数据序列,整理成5列录波图,并利用线性重投影算法形成可供神经网络识别的时序序列,使用加权卷积多列神经网络进行挖掘,将用户每1s步长的用电量信息分解成空调、插座、照明等负荷用电量信息,最终形成家庭智能用电管理系统的数据分解展示功能。经过与针对上述负荷单独安装电能表的实测数据进行对比,发现该改进智能用电优化算法得到的用电量分解结果,与实测结果的误差率均为5.8~6.0%之间。
In order to improve the calculation efficiency of the home intelligent power consumption management system,the current and voltage data sequences of 1s steps collected by the remote meter reading real-time fee controlled electric energy meter are sorted into 5-column oscillograms by using the time sequence analysis algorithm,and the linear re projection algorithm is used to form the time sequence that can be recognized by the neural network,and the weighted convolution multi column neural network is used for mining,The power consumption information of users per 1s step is decomposed into load power consumption information such as air conditioning,socket and lighting,and finally the data decomposition and display function of home intelligent power consumption management system is formed.By comparing with the measured data of the electric energy meter installed separately for the above load,it is found that the error rate between the power consumption decomposition results obtained by the improved intelligent power consumption optimization algorithm and the measured results is 5.8~6.0%.
作者
郭伟
迪里达尔·库尔班
姬彦君
Guo Wei;Dilidaer Kuerban;Ji Yanjun(State Grid Xinjiang Marketing Service Centre,Urumqi Xinjiang 830000,China)
出处
《现代科学仪器》
2022年第1期151-155,共5页
Modern Scientific Instruments
关键词
用电优化算法
家庭智能用电管理系统
神经网络算法
时序分析算法
用电负荷分解
Power Optimization Algorithm
Home Intelligent Power Management System
Neural Network Algorithm
Time Series Analysis Algorithm
Power Load Decomposition
