To address the limitations of traditional stability period discrimination methods, which suffer from insufficient discriminative information and difficulty in accurately identifying helicopter takeoff and landing states under complex wind and wave conditions, a multivariate hybrid model is proposed. This model integrates parallel bidirectional prediction branches with helicopter dynamics constraints, optimizing the utilization of environmental data. It transforms ship motion and wind field prediction results into helicopter takeoff and landing features. The model is built on a P-BiLSTM framework, employing a parallel bidirectional recursive structure to achieve synchronized forecasting of ship motion envelopes and wind speed time series. This forms an environment-motion information fusion prediction framework. Additionally, the dynamic constraint link rapidly computes the relative attitude and maneuvering state between the ship and helicopter, providing accurate state information for shipborne helicopter operations. Experimental data and simulation results demonstrate that the model achieves high prediction accuracy in wind and wave environments, offering effective support for takeoff and landing decision making.