Abstract— This paper focuses on assessing the behavior of the Unmanned Aerial Vehicle (UAV) through its previous flights as a response to an incident. The technique proposed in this paper helps in determining the abnormal flights, and the contribution of the variables in potential faults, in order to ensure the UAV safety. We statistically represent the behavior of the UAV through a flight by using the values of three features: The values of the Pearson Correlation Coefficient, the Y-Intercept, and the slope of the linear regression for each pair of the UAV variables; then, (Principal Components Analysis) PCA-based anomaly detector is used to extract the abnormal flights and the contributed variables in the potential faults. To test the algorithm’s efficiency, we used the MKAD synthetic dataset (Multiple Kernel based Anomaly Detection). This dataset is published for public use and includes discrete and continuous variables, which are previously injected by different types of faults. The conducted experiments showed similar results as the results of the well-known MKAD algorithm, where our approach detected 100% of the abnormal flights, with no false alarms. The advantage of our algorithm is that it is an unsupervised algorithm, so it did not require the massive training dataset as the MKAD method did.