Abstract: The control parameters can not be provided for the model-based control algorithm, which leads to the undesirable control performances of the system since the depth motion model parameters regarding the unmanned semi-submersible vehicles (USVs ) BQ-01 are unknown. For this purpose, a multi-identification model-based dynamic sliding mode control algorithm is presented for the investigation of system's control problem on the desired depth. In the proposed algorithm, the average-fitting-error method is used to reduce the excessive redundant model parameters, and thus the optimum model parameters is provided for the control algorithm by switching methods. And the state feedback method is used to bring about the chattering exponential decay of sliding mode controller so as to reduce the settling time. The experimental results of lake trials demonstrates that the proposed algorithm could offer the best model parameters for sliding mode control, and the dynamic sliding mode control with multi-identification models is capable of ensuring BQ-01 system to achieve the ideal control performance. Key

Key words: controlscienceandtechnology, unmannedsemi-submersiblevehicle, multipleidentificationmodel, dynamicslidingmodecontrol, controlperformance