Hard Landing Pattern Recognition and Precaution With QAR Data by Functional Data Analysis
Yan Zhong, Tong Liu, Fang Fang, Jia Ge, Bohao Xu, Xinbin Zhao
Abstract
Hard landing is one of the most common safety events in the aviation industry, which has been a critical concern of airlines and aviation administration for a long time. Although the analysis of Quick Access Recorder (QAR) data has the potential to illuminate the formation reason of a hard landing event, most existing methodologies overlook the curve characteristics of QAR parameters and focus on a straightforward prediction problem for hard landing. These methods usually lack interpretability and provide limited preventative insights. This paper presents the Hard Landing Pattern Recognition and Precaution Pipeline (HL3P), an innovative framework designed to recognize different landing patterns of flights and provide proactive suggestions against hard landing. Utilizing functional data analysis techniques, we first identify the key QAR parameters that have critical impacts on hard landing and subsequently recognize distinctive landing patterns that exhibit noticeable disparities. Through a detailed comparison of landing curves and pilot operations between normal and hard landing flights, we provide insights into the formation reason for hard landing and offer practicable landing advice for pilots.