基于卷积注意力的非侵入式负荷辨识算法

Non-intrusive Load Identification Algorithm Based on Convolutional Attention

针对目前非侵入式负荷监测算法准确率不高、训练耗时较长等问题,提出了一种基于卷积注意力的非侵入式负荷辨识算法。首先,对负荷数据设置最短运行和最短停止时长以降低测量误差带来的干扰。然后利用卷积神经网络对负荷数据进行训练,构建的神经网络包括编码器、时间池化器、解码器,并在解码器中引入卷积注意力模块来计算时间序列中当前时刻最重要的信息。最后利用UKDALE数据集对所提负荷辨识模型进行验证,并与现有算法进行对比。仿真结果表明,所提算法具有更好的辨识精度和泛化能力,训练所用时间减少约27.9%。

Aiming at the problems of low accuracy and long training time of traditional non-invasive load monitoring algorithm,this paper proposes a non-intrusive load identification algorithm based on convolutional attention.Firstly,the minimum running and stopping time are set for the load data to reduce the interference caused by measurement errors.Then the convolutional neural network is used to train the load data.The constructed neural network includes encoder,time pool converter and decoder,and the convolutional attention module is introduced into the decoder to calculate the most important information at the current moment in the time series.Finally,the proposed load identification model is verified by using the UKDALE dataset,and compared with the existing algorithms.Simulation results show that the proposed algorithm has better identification accuracy and generalization ability,and the training time is reduced by about 27.9%.