Fault Prediction Method of Access Control Terminal Based on Euclidean Distance Center SMOTE Method

Abstract

Access control fault data is an imbalanced dataset with noise. Current algorithms such as SMOTE (Synthetic Minority Oversampling Technique) and BSMOTE (Borderline-SMOTE) have defects in processing imbalanced dataset with noise. An improved SMOTE algorithm based on the Euclidean distance center, namely, CBSMOTE was proposed. CBSMOTE algorithm calculates the Euclidean distance center of the whole minority(abnormal) class samples, firstly, the noise and irrelevant samples will be removed according the center, remaining the appropriate samples. Secondly, the SMOTE algorithm is used to generate synthetic-samples on remaining samples. The experimental data uses the access control fault data as sample dataset. The imbalanced ratios of dataset were set to 0.1 and 0.05, respectively, and each imbalanced was grouped with 0.1, 0.2, and 0.3 noise ratios. The experimental results show that the performance of CBSMOTE algorithm over ROC, AUC, Accuracy, Precision, Recall and F1Score are better than that of BSMOTE algorithm and SMOTE algorithm, which effectively improves the classification accuracy of minority class samples under noise, thus improves the accuracy of access control terminal fault prediction.