Neutron Signal Denoising using Edge Preserving Kernel Regression Filter

끝점 신호 보존을 위한 적응 커널 필터를 이용한 중성자 신호 잡음 제거

A kernel regression filter with adaptive bandwidth is developed and successfully applied to digital reactivity meter for neutron signal measurement in nuclear reactors. The purpose of this work is not only reduction of the measurement noise but also the edge preservation of the reactivity signal. The performance of the filtering algorithm is demonstrated comparing with well known smoothing methods of conventional low-pass and bilateral filters. The developed method gives satisfactory filtering performance and edge preservation capability.