Real-time Angle Outlier Detection Based on Robust Preprocessing and Feature Extraction

For the traditional algorithm does not apply well in high-dimensional data stream and detecting potential outliers timely in the case of traditional outlier detection, A robust preprocessing and feature extraction method for real time angle based outlier detection algorithm is proposed to improve stability of irregular data sets and high dimensional spaces of the traditional algorithm. First, k-nearest neighbor and similarity method are applied for data clustering analysis. Then using angle vector to extract the key feature to lessen the data dimension, improving the computation speed of outlier detection algorithm in high dimensional space. Finally, the method based on outlier detection of angular distribution is used to identify outliers in high-dimensional data streams. The experimental results in real-world data sets corroborated that the suggested approach, which can improve the robustness of the algorithm in high-dimensional space, will optimize the storage space and reduce the time cost of the algorithm without loss of accuracy case, and provides a theoretical basis for the rapid detection of high dimensional outliers in the battery system model.