A MULTI-CLASSIFICATION IDENTIFICATION MODEL OF UAV FLIGHT RISKS BASED ON KMEANS AND XGBOOST

This paper presents a multi-classification identification model of UAV flight risk using K-means and XGBoost. With the advancements in unmanned aerial vehicle (UAV) technology, these vehicles are being increasingly utilized in various fields. At the same time, the research on anomaly detection using UAV flight parameter data to improve UAV safety has become one of the hot topics. Firstly, the study analyses two types of data-driven anomaly detection methods based on classification and similarity. Secondly, combining the ideas of the above two types of methods, we propose a multi-classification identification model of UAV flight risk based on K-means and XGBoost. Then, the experiments were conducted on a certain type of multi-rotor UAV flight data, demonstrating that the proposed model achieved a classification accuracy of 95.47%, precision of 95.52%, recall rate of 95.47%, and F1-score of 95.44%. Furthermore, the proposed model outperformed other machine learning models, such as Decision Tree (DT), SVM, and KNN, with a superior classification effect. These results highlight the effectiveness of the proposed approach for UAV flight risk identification, and its potential to enhance the safety and reliability of UAVs.