日本原子力学会

The extreme conditions in the reactor pressure vessels at the Fukushima Daiichi nuclear power plant, coupled with the reliance on remote sensors for decommissioning works, mean that most of the work will be carried out under a state of uncertainty. There is a need for a fast and rational framework for the treatment of measurement data from a multitude of sources in order to understand the condition inside the reactors, particularly in differentiating between statistical fluctuations, detector errors, and abnormal conditions. This paper proposes the use of Bayesian updating to produce quantitative probability estimates for all considered scenarios, which can be updated at the same time as new measurement data is collected. A simple model is used and several situations are considered in this paper to demonstrate the robustness of the Bayesian approach.