提出一种基于LSTM混合算法的用户电量数据异常检测模型。首先,提出了一种用于电力数据特征提取的检测器,该检测器通过从样本中提取抽象特征重新构造输入,并将重构误差与阈值进行比较,实现电力用户高维特征表示。其次,提出了混合CNNLSTM的电力数据异常分析网络,利用CNN叠加特征表示,并基于双层LSTM捕获电力数据上下文关系,从而有效提高模型的分类和回归能力。实验阶段,以某电力公司提供的电力数据为例,对所提模型进行验证。实验结果表明,所提模型性能最优,准确率和召回率分别为89.3%和69%。

In the paper,an anomaly detection model of user power data based on LSTM hybrid algorithm is proposed based on the study of CNN and LSTM.Firstly,a detector for feature extraction of power data is proposed.The detector extracts abstract features from samples,reconstructs the input,and compares the reconstruction error with the threshold to realize the high-dimensional feature representation of power users.Secondly,this paper proposes a hybrid cnn-lstm power data anomaly analysis network,which uses CNN superposition feature representation,and captures the context of power data based on double-layer LSTM,so as to effectively improve the classification and regression ability of the model.In the experimental stage,taking the power data provided by a power company as an example,the proposed model is verified.The results show that the performance of the proposed model is the best,and the accuracy and recall are 89.3% and 69% respectively.