Accurately predicting the touchdown speed of civil aviation aircraft is crucial for detecting unstable approaches and predicting runway occupancy time. This study proposes a CNN-BiLSTM-Attention hybrid architecture optimized by IGJO (Improved Golden Jackal Optimization) algorithm to achieve superior prediction accuracy. Considering the characteristic that the touchdown speed is affected by the coupling of multi - dimensional factors such as aircraft performance, approach status, meteorological conditions, and runway information, Initially, the convolutional neural network (CNN) extracts complex spatial features from the constructed data. Subsequently, a bidirectional long short-term memory (Bi-LSTM) network is employed to examine temporal relationships among these features. Attention mechanism module is incorporated to highlight important feature details using variable weighting. Finally, to tackle the problem of insufficient optimization ability of traditional hyperparameter optimization algorithms and fully optimize the hyperparameters of the above - mentioned hybrid network, the IGJO (improved golden jackal optimization algorithm) is utilized as a optimizer. By introducing the Gaussian random walk, spiral search, sine - cosine search strategy, and lens imaging opposition - based learning strategy, the convergence accuracy and global search ability of the GJO algorithm are significantly enhanced. The IGJO-optimized CNN-BiLSTM-ATTENTION model demonstrates superior prediction accuracy. Experimental findings indicate that the MAE, RMSE, and MAPE of its predictions are 3.2017, 3.06%, and 3.8817, correspondingly, which are more accurate than those of the unoptimized model and the models optimized by GJO, PSO, and DA. This prediction method provides strong support for air traffic control departments to accurately predict the touchdown speed.