New Approaches to Inverse Problems in Radiation Measurement

Abstract

  Approaches to inverse problems in radiation measurement are discussed specifically on the spectrum analyses and unfolding problems. A new technique based on the linear associative neural network for radiation spectrum analysis is first introduced. The technique was succesfully applied to complex gamma-ray spectra by NaI(Tl) detectors, and to highly overlapped X-ray spectra by Si(Li) for the PIXE analysis. Secondly, a new unfolding method by using the Bayes' theory alone is demonstrated for the neutron spectra observed by an NE213 spectrometer. Finally the emphasis is placed on the cooperative works between the sudies on 'direct' and 'inverse' problems in radiation measurement.