As air traffic has become increasingly congested， the effective mining of abnormal flight behaviors can assist controllers to make decisions. Current methods can only identify abnormal spatial positions with limited horizontal scalability. This paper incorporates four abnormal characteristics， position， speed， altitude and heading， and improves the density-based spatial clustering of applications with noise （DBSCAN） by the altitude level division strategy， the local outlier factor and the faster cover tree. Then， the density clustering algorithm named density-based spatial clustering considering speed， direction and high level improved by local outlier factor （LOFDBSC-SDH） is proposed to effectively extract the normal track patterns of airplanes. Next， based on the normal track patterns， a track matching algorithm is designed to mine abnormal flight behavior patterns， including flights’ cross-point time and above abnormal features. Finally， the effectiveness of the proposed scheme is verified by the experimental simulation.