Comparison of corellation method and artifitial neural networks for determining of object position by ultrawideband fields

Author:

Persanov I. D.1ORCID,Dumin O.M.1ORCID,Plakhtii V. A.1ORCID,Pryshchenko O. A.1ORCID,Fomin F. G.1ORCID

Affiliation:

1. V.N. Karazin Kharkiv National University, Kharkiv, Ukraine

Abstract

Background: Global and local positioning systems have a wide area of civil and military applications. Transport, logistics, precise agriculture, industrial technologies, safety systems need a strict definition of objects position on plane or in space. Existing modern positioning systems have some drawbacks in utilization and restrictions in application. Objectives: To improve a system of local positioning on a plane that does not need a time synchronization using the impulse ultrawideband electromagnetic field of two spaced bow-tie antennas and analysis and recognition of time forms of received waves by artificial neural networks and cross correlation method. To carried out the investigation of stability of the positioning system operation in presence of an interference in the form of additive white noise. Materials and methods: The electromagnetic simulation of excitation and radiation of the antennas is carried out by finite difference time domain method. The classification of received impulse form by known samples is realized by two alternative method, i.e. correlation approach and artificial neural networks. Results: The utilization of correlation method and artificial neural networks permitted to realize the positioning system with angular resolution of 1 degree. The probability distributions of recognized angles for different levels of additive noise in received signals for these two techniques are obtained. Conclusion: The comparison of artificial neural network application and correlation method for angle recognition shows that artificial neural networks can demonstrate a better precision than correlation approach. artificial neural network gives a correct angle recognition after statistical averaging of classification results even for the signal to noise ratio 0 dB. Artificial neural networks demonstrate a in three times shorter time of numerical simulation than we need for cross correlation function calculation. The application of shorter electromagnetic impulse increases the quality of angle classification in presence of the noise for both presented methods.

Publisher

V. N. Karazin Kharkiv National University

Subject

General Medicine

Cited by 1 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3