As a crucial flight data analysis technique, aircraft Flight Maneuver Recognition (FMR) aims to identify and classify various maneuvers performed by an aircraft during flight. This holds significant significance for flight safety, flight control, and flight performance evaluation. However, the precise classification and differentiation of various complex maneuvers in different flight scenarios and conditions remain a formidable task. In addition, aircraft maneuvers involve multiple aspects of information, such as position, velocity, attitude, etc. The importance differences of different feature channels involved in various aircraft maneuvers have often been ignored in previous studies. To address these issues, we propose a novel deep learning framework for aircraft trajectory recognition based on Efficient Channel Attention mechanism and Convolutional Long Short-Term Memory network (ECA-CLSTM) to enhance the model's robustness and weighting capabilities. The comparative experimental results show that our ECA-CLSTM model had the more recognition performance and noise robustness compared to using CNNs or LSTM models individually. This deep learning model provides a method and reference for flight maneuver recognition using sequence data.