Research On Data Fusion And Prediction Method For UUV Docking

We usually need Unmanned Underwater Vehicle(UUV) works for a very long timeunderwater. And its working time mostly restricted to the energy its takes. To accomplishenergy support, data recovery and new missions download, it is necessary to recover UUVsafely and effectively. The UUV underwater autonomous docking back which based on thedata fusion of multi-sensor can effectively shorten the UUV underwater time, improveefficiency and reduce the risk of recovery.During the docking process of the UUV, the underwater docking environment should beperceived.It mainly include the navigation positioning of itself and the relative dockingplatform. The positioning system is constructed of different kinds of sensors that UUV takes，including acoustics positioning and visual positioning system。Because of the two positioningsystem works in different scope and have different accuracy and reliability, to recover UUVsimply，effectively，and safely，this paper designs SBL navigation positioning system.Before data fusion, the positioning system needs to pre-treat sensors’ data. The systemalso put time and space calibration according to space coordinates’ disunity resulted bysensors’ installation and the opening time of sensors, sampling frequency respectively. Outlierremoved algorithms is designed based on data transformation frequency to deal withuncommon data of sensors.According to UUV autonomic recovery process, a short baseline and visual positionfusion system mode is established, and studied the covariance matching based on fuzzyadaptive data fusion algorithms, and focus on the Kalman prediction method for multi-steprecovery movement UUV location information for a very short-term forecast.Finally, through the loop simulation of the short baseline data and visual experimentalspace and time calibration and outlier rejection and filtering algorithm validation, and with theoff-line wavelet denoising compared the effects of verification, confirmed by the data rate ofchange algorithms and Kalman filtering outliers, the experimental data can provide the fusionsystem a good system input. Through covariance matching fuzzy adaptive fusion to shortbaseline positioning and visual guidance system experiment, can provide the control system amore reliable control information Kalman based prediction of multi-step simulation exerciseshows that in the very near future to give some precision advance control signal.