Enhanced Accuracy Of Optical Target Positioning Of UAV Based On Error Identification And Compensation

As an advanced combat weapon,unmanned reconnaissance aircraft(UCAV)has appeared frequently in several local wars and has repeatedly made great achievements.It has Attracted wide attention in the world.The UAV positioning system is one of the most basic task execution units of the unmanned reconnaissance aircraft,and its positioning accuracy directly affects the operational performance of the UAV.Optical target positioning is one of the commonly used positioning methods for drones.It calculates the position information of the target by the laser ranging value,the attitude angle of the drone and the pointing angle of the photoelectric reconnaissance platform.However,the error of the measured value is inevitably Affects the accuracy of target positioning.This article has carried out a series of work on how to improve the positioning accuracy of drones,including:(1)This paper first studies the target positioning principle of UAV.The next job is to establishe the target positioning error model based on MATLAB platform,and uses Monte-Carlo method to analyze each error factor separately.This paper has explained the influence degree of each error on the positioning accuracy and Laid the theoretical foundation for the next work.(2)In this paper,the attitude angle error and the pointing angle error that affect the positioning accuracy of the UAV are equivalent to the line-of-sight pointing offset angle.The problem of solving the error terms of the UAV positioning is transformed into the solution of the offset angle.Finally,the error compensation model of the UAV positioning system is established.This paper also designed a scientific flight calibration scheme,which can identify the error parameters of UAV more effectively.(3)Aiming at the error factors that are difficult to be represented by the actual positioning,based on the basic parameter model,an improved algorithm based on semi-parametric regression model is proposed.The simulation proves that the algorithm can effectively suppress non-parametric error interference and further improve the positioning.Accuracy makes the system error compensation model more suitable for practical applications.In this paper,wavelet threshold denoising and BP neural network methods are applied to the extraction of nonlinear system errors,and the effectiveness of the method is verified by simulation.