Reactor condition monitoring and singularity detection via wavelet and use of entropy in monte carlo calculation

Abstract

Wavelet theory was applied to detect the singularity in reactor power signal. Compared to Fourier transform, wavelet transform has localization properties in space and frequency. Therefore, by wavelet transform after de-noising, singular points can be found easily. To demonstrate this, we generated reactor power signals using a HANARO (a Korean multi-purpose research reactor) dynamics model consisting of 39 nonlinear differential equations and Gaussian noise. We applied wavelet transform decomposition and de-noising procedures to these signals. It was effective to detect the singular events such as sudden reactivity change and abrupt intrinsic property changes. Thus this method could be profitably utilized in a real-time system for automatic event recognition (e.g., reactor condition monitoring). In addition, using the wavelet de-noising concept, variance reduction of Monte Carlo result was tried. To get correct solution in Monte Carlo calculation, small uncertainty is required and it is quite time-consuming on a computer. Instead of long-time calculation in the Monte Carlo code (MCNP), wavelet de-noising can be performed to get small uncertainties. We applied this idea to MCNP results of $k_{eff}$ and fission source. Variance was reduced somewhat while the average value is kept constant. In MCNP criticality calculation, initial guess for the fission distribution is used and it could give contamination to solution. To avoid this situation, sufficient number of initial generations should be discarded, and they are called inactive cycles. Convergence check can give guildeline to determine when we should start the active cycles. Various entropy functions are tried to check the convergence of fission distribution. Some entropy functions reflect the convergence behavior of fission distribution well. Entropy could be a powerful method to determine inactive/active cycles in MCNP calculation.