System identification of a research nuclear reactor versus loss of flow accident using recurrent neural network
by Behzad Salmasian; Gholam Reza Ansarifar; Seyeed Mohammad Mirvakili
International Journal of Nuclear Energy Science and Technology (IJNEST), Vol. 12, No. 3, 2018

Abstract: In this paper, modelling of the Tehran Research Reactor is done using Recurrent Neural Network (RNN) in Loss of Flow Accident (LOFA). TRANS code is calculated as training data mode for each of the scenarios. Supervised recurrent neural network is chosen for modelling and identification system, classified system data and appropriate parameters for modelling function of system have been chosen, then data is classified. In the next step, we choose variant networks to train and compare with each other. Next, an optimised network is chosen according to mean square error parameter and correlation among educational data from TRANS code and network output data. Finally, entrance data related to the unforeseen accident was entered to the system and the predicted results by model and output data of TRANS code were compared. Results demonstrate the appropriate conformity between extraction data of TRANS code and extraction data of the model, which shows appropriate function of the model.

Online publication date: Tue, 16-Oct-2018

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Nuclear Energy Science and Technology (IJNEST):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com