An Efficient Method Combined Data-Driven for Detecting Electricity Theft with Stacking Structure Based on Grey Relation Analysis

Nowadays, electricity theft has been a major problem worldwide. Although many single-classification algorithms or an ensemble of single learners (i.e., homogeneous learning) have proven able to automatically identify suspicious customers in recent years, after the accuracy these methods reaches certain level, it still cannot be improved even if continues optimized. To break through this bottleneck, heterogeneous learning method with stacking integrated structure different strong individual for detection is presented paper. Firstly, we use grey relation analysis (GRA) select classifier combination LG + LSTM KNN as base model layer based on principle highest comprehensive evaluation index value. Secondly, support vector machine (SVM) relatively good results overall experiment selected meta-model layer. In way, constructed. Finally, experiments are conducted consumption data from State Grid Corporation China, and show that performance proposed better than existing state-of-the-art (where area under receiver operating characteristic curve (AUC) value 0.98675).