Research On Error Calibration Method Of UUV Navigation Aided By UTP

Navigation and positioning is the guarantee for Underwater Unmanned Vehicle(UUV)finishing tasks.The precision of navigation and positioning has an important influence on the UUV operating performance.When UUV perform underwater operation,or in other GPS failure situations,we can only rely on the traditional navigation and positioning method such as Dead Reckoning(DR)or Strapdown Inertial Navigation System(SINS),etc.However,due to their working mechanism of these traditional navigation and positioning methods,the navigation error will accumulate when an UUV performs long distance voyage underwater.Therefore,the acoustic auxiliary navigation and positioning method using Underwater Transponder Positioning(UTP)is introduced to calibrate the navigation error of an UUV,thus improving the navigation and positioning accuracy.In UTP auxiliary UUV navigation Error Calibration procedure,UTP workflow,calculation method,and the observability are the foundation of positioning,and the calibration of transponder coordinate and system error analysis are the premise to reduce the positioning error,and the algorithm to suppress filtering divergence and multi-sensor integrated navigation algorithm are the guarantee to improve the navigation and positioning performance.The essay focuses on the study of the above aspects,and the main contents are as follows:For the implementation issues of UTP navigation positioning,this paper introduces the principle,composition,and calculating methods of UTP,and analyzes in detail observability conditions of UTP navigation based on the nonlinear system observability criterion,and then uses Cubature Kalman Filter(CKF)to solve UTP.For the calibration problem of transponder coordinates,an improved particle swarm optimization(PSO)algorithm is introduced.In the process of evolution,each particle evolves along different directions respectively and generates two homologous particles.Then the two particles are compared and the better one is kept for the next generation.At the same time,an adaptive adjustment algorithm of learning factor is introduced to guarantee the diversity of the search.The improved algorithm can obtain higher precision for transponder coordinates.Then the influence of sensor installation error,measuring error,sound ray bending,and floating of transceiver on the positioning precision are analyzed.For the filtering divergence suppression problem,the calculation error,the inaccuracy of the observation noise model,and the inaccuracy of the system state model are the three aspects in the filter designation.The covariance matrix of errors will probably be nonnegative due to calculation error,and in order to solve this we perform square root decomposition to the covariance matrix.As to the inaccurate problem of the observation noise model,we estimate the covariance of the observation noise based on the variable Bayesian inference principle,and then propose a kind of adjustment method for variational adjustment factor to ensure the filtering performance.Then variable Bayesian inference principle is used to directly estimate the covariance,thus obtaining higher filtering accuracy.In view of the inaccuracy of the system state model,variable Bayesian inference state covariance estimate algorithm is proposed.In view of the inaccuracy of both the observation noise model and the system state model,the covariance matrices of noise and state estimated by variable Bayesian inference are combined to form variable Bayesian hybrid filter,and then convergence criterion is used to switch between the two situations.As to multi-sensor data fusion problem,the asynchronous multi-scale sequential filtering algorithm is proposed.This algorithm uses asynchronous sequential block structure form and the wavelet multi-scale transform to decompose the signals into different scales,with fusion filtering carried out respectively,and then the wavelet reconstruction is performed.In order to further improve the filtering precision,the weighted asynchronous multi-scale sequential filtering algorithm is proposed.In view of the filtering divergence problem,variable Bayesian hybrid filter is introduced into asynchronous multi-scale sequential weighted filtering algorithm.The simulation is carried out to verify the effectiveness of the proposed algorithm.