FMCW 전파고도계의 딥러닝 기반 고도 추정 (Deep Learning-Based Altitude Estimation for an FMCW Radar Altimeter)

서지정보

· 발행기관 : 한국전자파학회
· 수록지 정보 : 한국전자파학회 논문지 / 36권 / 2호 / 202 ~ 211페이지
· 저자명 : 장종훈, 안병철

초록

본 논문에서는 딥러닝에 기반하여 주파수 변조 연속파(FMCW) 전파고도계가 고도를 추정하는 방법을 제시한다.
FMCW 전파고도계에인공신경망을도입하고, 다양한지형에서수집한비트주파수스펙트럼과직하방고도를입력하여인공신경망을학습한다. 전파고도계는임의의비트주파수스펙트럼을학습된인공신경망에입력받아, 각비트주파수가직하방고도일확률분포를출력함으로써고도를추정한다. 비행시험을통해수집한데이터를이용해고도오차를비교⋅분석한결과딥러닝방법으로추정한고도가종래의고도추정방법대비고도오차가평균적으로적어지는결과를얻을 수있었다.

영어초록

This study presents a method for estimating altitude using a frequency-modulated continuous wave (FMCW) radar altimeter based on deep learning (DL). An artificial neural network (ANN) was implemented in the FMCW radar altimeter, which was trained using beat-frequency spectra and nadir altitudes collected from various terrain. The radar altimeter estimates the altitude by inputting an arbi trary beat frequency spectrum into the trained ANN and outputting a probability distribution of each beat frequency corresponding to the nadir altitude. The altitude errors of the conventional and DL-based methods were compared and analyzed using flight data. On average, the altitude estimated by the DL-based method had a smaller error than that estimated by the conventional method.

참고자료

· 없음