<p>Due to complex environmental disturbances, high-precision tidal prediction remains a significant challenge in ocean engineering applications. To address tidal variations characterized by nonlinearity, uncertainty, and time-varying dynamics, a hybrid prediction scheme incorporating adaptive module adjustment (AMA) is proposed. The harmonic analysis method is first applied to model tidal effects induced by the movements of celestial bodies. Subsequently, residual components are decomposed using empirical mode decomposition (EMD), with long short-term memory (LSTM) networks and polynomial fitting (PF) employed to construct the tidal prediction model. The decomposition order for LSTM input time series and the selection of polynomial modules are determined adaptively. Finally, the predictions from harmonic analysis and the ensemble model components are combined to generate the final tidal forecast. Experimental simulations are conducted using observed tidal data from gauges at Canaveral Port and Old Port Tampa. Simulation results demonstrate that the proposed adaptive tidal prediction model outperforms conventional methods in terms of prediction accuracy.</p>