In order to overcome the issues of thrust saturation, uneven energy consumption and suboptimal control associated with the traditional hierarchical control strategy employed in ship dynamic positioning (DP) systems, an integrative control allocation strategy is proposed by combining the low-level thrust allocation into the high-level motion controller. The proposed scheme is implemented via the neural network-based nonlinear predictive control (NNPC) and solved via an improved particle swarm optimization (PSO) algorithm. A data-driven recurrent neural network (RNN) ship motion and thrust allocation model is off-line identified based only on the current and historical measured output and thrust input data. Utilizing the established RNN model to predict the future multi-step outputs, and combining the motion control, the thruster energy consumption and the thruster wear and tear as an integrative control objective, the integrated NNPC online optimization problem with thruster constraints is established and solved by an improved evolution algorithm of PSO. Simulations are respectively carried out for setpoint regulation and trajectory tracking. Simulation results well demonstrate the feasibility and superiority of the proposed control strategy.