Inverse estimation of unknown radioactive source using detection probability and adjoint calculation

In order to investigate distribution of fuel debris remaining in the reactor containment vessel of Fukushima Daiichi NPS, we focused on the estimations based on the measured value of neutron counts. Maximum Likelihood-Expectation Maximization (ML-EM) method and Moore-Penrose Matrix Inverse (MPMI) method were examined. The ML-EM method is used for image reconstruction of Computed Tomography, and the MPMI method is one of the solution of the simultaneous linear equations. Concrete shielding were set between detectors and neutron sources in the calculation system, since where detector position in actual applications would be outside of the shielding such as the pedestal. Since sufficient number of count measurement positions would not be secured owing to the complexity of structures inside containment vessel, the number of measurement points were set to be smaller than that of neutron sources. Since the detection probability is used as the coupling coefficient between the radiation count and the radioactivity, the adjoint transport calculation is performed to obtain the detection probability. The detection probability calculated by adjoint neutron flux reproduced the physical trend expected from detector and neutron source positions. Result of estimation using this detection probability indicates that two methods gave reasonable neutron emission rate distribution that is close to the true value. (authors)