Key Performance Upgrading Technology Research For Super-mini Underwater Vehicles

Super-miniUnderwater Vehicles(SUV) are widely used in the shallow waters of rivers and lakes, they have now been widely used in civil and military fields such as hidden defectsdetection of dam and dyke,underwaterstructure repair and examination of offshore drilling platforms,repairing of submarine optical cable,kelp reaping,green algaeexploration,laying and sweeping of mines,underwater reconnaissance and so on.The paper aims to study the SUV designment and its key technology, including pose detection, control system and image acquisition/processing system. Pose detection of the underwater vehicle plays a decisive role in its controlling and navigation. The control system, the foundational core of SUV,it decides the maneuverability and reliability of the underwater vehicle. The image acquisition/processing system are methods for sense underwater environment.The main work of this paper includes:(1) A fast intelligent positioning method for underwater vehicle. Common tool of underwater vehicle positioning system is Short BaseLine positioning sonar (SBL), which is low speed and unreliable. Based on the comparative analysis of usual probabilistic location algorithm of robots, the particle swarm optimization method is used to calculate the position of vehicle between measured SBL signals to speed up the positioning system.It fuses the SBL, X/Y angle sensor and depth sensor effectually to make up the hardware architecture of SUV positioning system, and develop the software of positioning system adopting component-based method. Positioning effect test experiment is carried out in a ship model tank, the results show that this method can improve the speed of underwater vehicle positioning system, and make SBL synchronized with pose sensors.(2) Scanning sonar image correction based on positioning system and morphology. The sonar data in single scan angle is processed by Recursive Least Squares (RLS) filter. The distortion of scanning sonar image brought in by posture change of ROV is corrected by the results detected by pose detection system proposed in the paper. Then the variable structure element of mathematical morphology operation method is used for the corrected scan sonar images for further processing.It can be seen from the results, binary side scanning sonar image can provide facilitate for the identification of obstacles(3) An open SUV controlling system based on subsumption architecture. The traditional hierarchical structure of the robot control system can not satisfy the demands of underwater vehicles because of complexity work environment and variability of underwater tasks, as to a simple open system, itmay cause deterioration or low efficiency. In result, this paper proposes a subsumption structure based open control system, which is roughly divided into several modules, integrate the modules based on subsumption structure while the subsystems'integration of each module based on open architecture. Analyse the main contents of underwater vechile controlling according to its working characteristics. A control scheme is applied in motion control,which is based on Q-Learning (QL) agent combined with Radial Basis Function (RBF) neural network control algorithm. The mean square error of the method used in this paper is smaller than pure RBF neural networkcontrol method in the heading angle lock simulation experiment.