A hybrid data-model-driven method is proposed for predicting tower crane displacement responses under extreme typhoon conditions. Using a tower crane from an actual engineering project as a case study, an IoT-based real-time monitoring system is established to obtain displacement response data throughout the construction period. In parallel, a finite element model is developed to simulate the structural displacement of the tower crane under extreme wind loads, compensating for the scarcity of extreme-condition samples in the monitoring data. Based on the CNN-BiLSTM-AdaBoost algorithm, both purely data-driven and hybrid data-model-driven displacement predictions are conducted for normal wind conditions and extreme typhoon conditions. The results indicate that purely data-driven methods exhibit limited predictive capability under extreme conditions and tend to underestimate structural responses. In contrast, the hybrid data-model-driven approach effectively integrates measured data with finite element simulation results, significantly improving the accuracy and reliability of displacement predictions under extreme typhoon loading.