ニューラルネットワークを用いたミリ波異物検査システムの研究

概要

論文の詳細を見る
Alien substance mixture accident inside foods is serious problems for both consumer and company side. The systems that can measure the different kinds of alien substance have been required. The detecting systems using X-ray, SQUID sensor, and supersonic wave as well as metal detector have been developed and been in the market, however, they are, in many cases, appropriate for the detection of metal alien substance. We have been studying the system using millimeter-waves with frequency of 50-140 GHz to detect both metal and dielectric alien substance. The scattered and transmitted waves are detected using a heterodyne receiver. The angular distribution of the scattered wave is analyzed to obtain the dispersion, auto-correlation function, and the Fourier spectrum. The neural network analysis is utilized to judge the mixture and the size of alien substances. It is shown that the alien substance having 1-3 mm size can be detected in spite of the materials by the present system. By using short wavelength millimeter wave, it is expected to detect even less than 1 mm alien substance.
九州大学大学院総合理工学府,Interdisciplinary Graduate School of Engineering Sciences, Kyushu Universityの論文