Determination of Reactivity and Neutron Flux Using Modified Neural Network for HTGR

• Main Authors: M. Subekti, K. Kudo, K. Nabeshima, K. Takamatsu
• Format: Article
• Language: English
• Published: Center for Development of Nuclear Informatics, National Nuclear Energy Agency (BATAN) 2017-08-01
• Series: Atom Indonesia
• Subjects: HTTR; Reactivity determination; Method development; Verivication; Withdrawal test; Online application
• Online Access: http://aij.batan.go.id/index.php/aij/article/view/683

<p class="MsoNormal" style="margin-top: 0cm; margin-right: -5.1pt; margin-bottom: 4.0pt; margin-left: -5.4pt; text-align: justify;"><span style="font-size: 9.0pt;">Nuclear kinetic calculations based on point kinetic model have been generally applied as the standard method for neutronics codes. As the central control rod (C-CR) withdrawal test has demonstrated in a prismatic core type high-temperature gas-cooled reactor (HTGR) named High Temperature Engineering Test Reactor (HTTR), the transient calculation of kinetic parameter, reactivity, and neutron fluxes, requires a new method to shorten calculation-process time. Development of neural network method was applied to point kinetic model as the necessity of real-time calculation that could work in parallel with the digital reactivity meter. The combination of </span><span style="font-size: 9.0pt; mso-ansi-language: IN;" lang="IN">Time Delayed Neural Network (</span><span style="font-size: 9.0pt;">TDNN</span><span style="font-size: 9.0pt; mso-ansi-language: IN;" lang="IN">)</span><span style="font-size: 9.0pt;"> and </span><span style="font-size: 9.0pt; mso-ansi-language: IN;" lang="IN">Jordan Recurrent Neural Network (</span><span style="font-size: 9.0pt;">Jordan RNN</span><span style="font-size: 9.0pt; mso-ansi-language: IN;" lang="IN">) named </span><span style="font-size: 9.0pt;">TD-Jordan RNN was the result of the modeling approach. The application of TD-Jordan RNN with adequate learning, tested offline, determined results accurately even when signal inputs were noisy. Furthermore, the preprocessing for neural network input utilized noise reduction as one of the equations to transform two of twelve time-delayed inputs into power corrected inputs.</span></p>