基于递归神经网络的智能电表运行误差远程估计方法

Remote estimation method of operation error of smart electricity meter based on recursive neural network

传统的电表运行误差估计方法难以控制时滞信息,导致其误差率偏高。针对该问题,该文提出了基于递归神经网络的智能电表运行误差远程估计方法。利用递归神经网络结构模拟高阶系统,通过学习网络连接方式获得最佳的解码效果,从而有效地控制延时;建立递归神经网络误差估计模型,得到估计结果;搭建智能电表运行误差远程检测系统,实现误差远程检测,并在此基础上,设计运行误差估计流程。由试验结果可知,该方法与2021年2月到5月份的实际智能电表数据统计结果一致,均在4月上旬误差率达到最低为-6%,证明了其估计结果较为精准。

The traditional error estimation method of electricity meter operation is difficult to control the time delay information,which leads to its high error rate.To solve this problem,this paper proposes a method of remote estimation of operation error of smart electricity meter based on recursive neural network.The recursive neural network structure is used to simulate the high order system,and the best decoding effect is obtained by learning the network connection mode,so the delay and lag can be controlled effectively.The recursive neural network error estimation model is established and the estimation results are obtained.The remote error detection system of smart electricity meter operation is built to realize the remote error detection.On this basis,the error estimation process of operation is designed.According to the test results,this method is consistent with the statistical results of the actual smart meter data from February to May 2021,with the error rate reaching the lowest-6%in the first ten days of April,which proves that its estimation results are accurate.