Research On Underwater Positioning Algorithm Based On Probability Graph Model

With the rapid development of acoustic technology,underwater positioning is playing an increasingly important role in the field of marine resource development and marine military fields.In order to accurately obtain the position of an underwater target,a long baseline positioning method is often used.However,in the long baseline positioning process,sparse beacons or lack of beacon communication information will lead to insufficient measurement information,so that the effective position of the target cannot be obtained.Therefore,the moving target needs to be combined with the measurement information of other positions and times,and the multi-parameter fusion estimation method is used to solve it,but it is difficult to obtain an analytical solution for the multi-parameter fusion.Aiming at the above two problems,this paper adopts the method of probabilistic graph model to estimate the target.This method uses the distance between the target and the node as the input to indicate the spatial interconnection,and the target velocity as the input indicates the interconnection between the positions of two adjacent moments.After the probability map is established,the sum-product algorithm is used to obtain the expression of the marginal function of the variable through the transmission and update of the message.By studying the principle of probabilistic graph model and underwater positioning technology,this paper establishes a factor graph model for three typical scenarios(absolute beacon calibration,relative beacon calibration,target positioning tracking)in the long baseline positioning process,and then uses the sum product algorithm to derive Get the marginal probability distribution function of the target location variable.Simulation analysis of the impact of resolution on positioning results.After the resolution exceeds the ranging error,the determining factor of the positioning error is the size of the ranging error,and the selection resolution should be smaller than the ranging error.Analyze the influence of the ranging error on the positioning result.When the random ranging error is the same,the node has the same influence on the positioning result,and the area with high positioning result accuracy is approximately a circle.With the increase of the ranging error,the positioning accuracy decreases,and the distribution is linear.The positioning results are compared with traditional methods.In the absolute calibration and relative calibration scenarios,the amount of data is sufficient,and the positioning accuracy of the probability map method and the least square method is the same.In the multi-beacon positioning process,the least square method only uses the current time data.The extended Kalman filter uses position recursion for positioning and tracking,and uses the position information of the previous moment and the current moment information,and the fusion information is less.The probability map method combines multiple time measurement information,and the probability map method has a higher positioning accuracy.In the single beacon positioning and tracking project,a single beacon is used,and the amount of measurement is insufficient,and the least square method cannot accurately locate at every moment.When the extended Kalman filter has no initial position,the trajectory cannot be estimated.When there is an initial position,the error is relatively large.The probability map combined with the ranging and speed information at multiple moments can describe the target trajectory well.Finally,this paper has passed the lake test verification,absolute calibration and relative calibration,sufficient data volume,and the probability map method and the least square method positioning accuracy are of the same order of magnitude.In the process of multi-beacon and single-beacon positioning and tracking,the position of the target at each moment can be estimated,and the trajectory of the target can be described.