Analysis and Research of Anti-Electric Theft Big Data Based on LSTM and RUSBoost
An anti-theft big data analysis model based on long short-term memory(LSTM)and random undersampling enhancement(RUSBoost)is proposed on a real-time sequence dataset.The used model consists of LSTM algorithm and RUSBoost technique.Normal-ization and interpolation methods are used to pre-process the electricity data to eliminate zero and undefined values.The relevant fea-tures are extracted from the preprocessed data by using LSTM algorithm for feature refinement of the data.Parameter optimization using classifiers in solving the electricity theft detection(ETD)problem can handle larger time series data.To enhance the performance of the RUSBoost method,the SVM,LR and CNN-LSTM models are compared using the bat algorithm for parameter optimization.Finally,the RUSBoost method is applied to balance the data effectively.The proposed model achieves an F1 score of 96.1%,an accuracy of 88.9%,a recall of 91.09%and a ROC-AUC score of 87.9%.All performance metrics aspects are better than the given conventional scheme.
LSTMRUSBoostanti-electric theftbig data analysiselectrical loss
国家科技期刊平台
NSTL
万方数据
