CAJ | 학술논문

目前三维空间测向定位多采用先在二维体系中计算然后转换为空间坐标的方法,正北偏差、投影变换的方位变形等会对定位精度造成很大的影响.本文立足于三维空间,利用多个测向平面形成的空间目标观测方程组,并利用实神经网络代数算法收敛速度快、时间效率高以及隐层神经元个数选取等方面的优势,构建了三站测向定位模型.仿真实验表明,本算法定位精度明显提高,在不同测向精度条件下定位误差接近CRLB,使得该算法适用于复杂的测向环境,具有很强的实用性.
목전삼유공간측향정위다채용선재이유체계중계산연후전환위공간좌표적방법,정북편차、투영변환적방위변형등회대정위정도조성흔대적영향.본문립족우삼유공간,이용다개측향평면형성적공간목표관측방정조,병이용실신경망락대수산법수렴속도쾌、시간효솔고이급은층신경원개수선취등방면적우세,구건료삼참측향정위모형.방진실험표명,본산법정위정도명현제고,재불동측향정도조건하정위오차접근CRLB,사득해산법괄용우복잡적측향배경,구유흔강적실용성.
At present, the three-dimensional spatial location is usually first calculated in two dimensional sys?tems, and then converted to the spatial coordinate. But north deviation azimuth deformation and projection trans?formation have a great effect on positioning precision. This paper uses the space targets observation equations formed with a plurality of direction finding plane, and then uses algebra algorithm of real neural networks with its fast convergence speed, high time efficiency and the advantage of number selection of neurons in hidden lay?er. So it is first applied to the 3D direction-finding, constructing direction finding and location model based on the three-dimensional space. Simulation results show that this algorithm positioning accuracy is obviously im?proved and that positioning error is close to CRLB under the condition of different direction finding precision, so the algorithm is applicable to complex measurement environment.