Research On Data Fusion Method Of Asynchronous Sensor System For Dynamic Positioning Ship

With the rise of more and more marine projects,the application of dynamic positioning ships has become more and more extensive.In the process of in-depth exploration and operation of marine resources,higher requirements are put forward for the sensor equipment equipped with dynamic positioning ships.Dynamic positioning ships use a variety of position sensors to provide corresponding positioning information for ship positioning.There are many interference factors in the measurement,which make the measured data deviate and affect the positioning accuracy of the ship.Multi-sensor data fusion can solve the corresponding problems.The data measured by each position sensor has problems such as noise and inconsistent sampling time.The data measured by multiple position sensors need to be filtered and time-synchronized,and then the position sensor data information is fused to provide an accurate position for the dynamic positioning ship information,thereby improving the positioning accuracy and overall performance of the dynamic positioning ship.The main research contents of this paper are as follows:Firstly,this article establishes the kinematics and dynamics model of the dynamic positioning ship according to the dynamic characteristics of the dynamic positioning ship;the measurement and positioning principle of the position sensor are analyzed,and the mathematical model of the differential global positioning system position sensor and the mathematical model of the underwater acoustic position sensor and the mathematical model of the tension cable position sensor are established;the influence of sea breeze,ocean wave and ocean current on the sensor is analyzed,and the marine environment interference force model is established.Secondly,in view of the noise interference in the data measured by the asynchronous sensor system of the dynamic positioning ship and the non-linear state of the measured data,the nonlinear filtering of the asynchronous sensor system of the dynamic positioning ship is studied.The importance density function of traditional Gaussian particle filter will reduce the filtering accuracy.This paper uses the seventh-order volume Kalman filter to reconstruct the importance density function,and uses the time update step of the seventh-order volume Kalman filter to improve the time lag of the particle filter.A method of combining Gaussian particle filter and seventh-order volumetric Kalman filter is proposed,which improves the accuracy of system filtering and reduces the influence of noise interference from the asynchronous sensor system of the dynamic positioning ship.Thirdly,to solve the problem of interpolation error in the time synchronization process of the dynamic positioning ship asynchronous sensor system,a Kalman-based asynchronous sensor system time registration method is proposed.For the dynamic positioning ship sensor,there is interpolation during the synchronization process.Error problem,the Kalman filter is introduced to filter the data after interpolation,which reduces the influence of interpolation error and solves the problem of errors in the interpolation process of the asynchronous sensor system of the dynamic positioning ship.Finally,in view of the deviation of the data measured by the asynchronous sensor system of the dynamic positioning ship,a multi-model adaptive deep belief network(M-ADBN)processing method is proposed,which uses the data of each state variable measured by the position sensor to establish a sub-adaptive deep belief network model,process the data of each state variable of the position sensor,and obtain a multi-dimensional state variable fusion model,complete the asynchronous sensor of the dynamic positioning ship.The data fusion processing of the multi-dimensional state of the system provides more accurate position information data for the dynamic positioning ship.