基于Hadoop分布式计算的混合神经网络负荷分类模型

为了对电力物联网背景下的海量负荷数据进行精细化分析,从中提取用电模式,提出一种基于Hadoop分布式并行计算的混合神经网络分类模型。首先,基于时间维度的一维卷积神经网络(1DCNN)搭建“负荷特征提取器”；其次,使用长短期记忆网络(LSTM)搭建“序列分类器”；最后,将该“混合神经网络分类方法”在Hadoop平台上搭建,实现算法的并行化运行,以适用于海量负荷曲线的高效辨识。使用标准时序数据与真实负荷数据测试该方法的分类性能,算例结果表明所提分类方法具有较高的分类精度,经并行化处理后有效提高了负荷数据的处理效率。

Hybrid Neural Network Load Classification Model Based on Hadoop Distributed Platform

In order to analysis the massive load data in the context of power internet of things, and extract the power mode, a hybrid neural network classification model based on Hadoop distributed parallel computing is proposed. Firstly, a one-dimensional convolutional neural network (1DCNN) base on time-dimensional is used to build the load feature extractor. Secondly, long-short-term memory network (LSTM) is used to build a sequence classifier. Finally, the hybrid neural network classification method is parallelized on the Hadoop platform for mass load classification. Standard time series data and real load data are used for analysis and test. The results of the examples show that the proposed classification method gainedSbetterSprecisionSofSclassification and improved load data processing efficiency after parallelization.