Neural control for power ascent of a TRIGA reactor

J. Humberto Pérez-Cruz, Alexander Poznyak

Research output: Contribution to conferencePaperResearch

5 Citations (Scopus)

Abstract

A basic control problem in a nuclear research reactor consists of increasing or decreasing the neutron power from a certain level R0 to a new desired level R1 and maintain the reactor stable at the new power level. For security reasons, this task must be performed in such a way that, during the power ascent, the instantaneous period of the reactor must always be greater than or equal to a lower limit value. To solve this problem, avoiding the difficulties associated with the physical modeling of the nuclear process, in this paper, we propose to use an indirect adaptive control scheme in which a single layer second order differential neural network achieves the on-line identification based only on three variables: the external reactivity, the fuel temperature, and the neutron power. The mathematical model provided by this identification process is employed to accomplish the control action in two stages. During the transient stage, the controller objective is to maintain the plant on a constant period. Once the desired power is reached, the control action is switched to a regulation stage. This identifier-controller is tested by simulation. Instead of the real plant, an eighth order physical model of a TRIGA reactor considered as a black box is used. The results show a good performance of the suggested approach. ©2008 AACC.
Original languageAmerican English
Pages2190-2195
Number of pages6
DOIs
StatePublished - 30 Sep 2008
Externally publishedYes
EventProceedings of the American Control Conference -
Duration: 30 Sep 2008 → …

Conference

ConferenceProceedings of the American Control Conference
Period30/09/08 → …

Fingerprint

Neutrons
Controllers
Research reactors
Mathematical models
Neural networks
Temperature

Cite this

Pérez-Cruz, J. H., & Poznyak, A. (2008). Neural control for power ascent of a TRIGA reactor. 2190-2195. Paper presented at Proceedings of the American Control Conference, . https://doi.org/10.1109/ACC.2008.4586817
Pérez-Cruz, J. Humberto ; Poznyak, Alexander. / Neural control for power ascent of a TRIGA reactor. Paper presented at Proceedings of the American Control Conference, .6 p.
@conference{1fc742e01d4f4234911f87f65bbbda3e,
title = "Neural control for power ascent of a TRIGA reactor",
abstract = "A basic control problem in a nuclear research reactor consists of increasing or decreasing the neutron power from a certain level R0 to a new desired level R1 and maintain the reactor stable at the new power level. For security reasons, this task must be performed in such a way that, during the power ascent, the instantaneous period of the reactor must always be greater than or equal to a lower limit value. To solve this problem, avoiding the difficulties associated with the physical modeling of the nuclear process, in this paper, we propose to use an indirect adaptive control scheme in which a single layer second order differential neural network achieves the on-line identification based only on three variables: the external reactivity, the fuel temperature, and the neutron power. The mathematical model provided by this identification process is employed to accomplish the control action in two stages. During the transient stage, the controller objective is to maintain the plant on a constant period. Once the desired power is reached, the control action is switched to a regulation stage. This identifier-controller is tested by simulation. Instead of the real plant, an eighth order physical model of a TRIGA reactor considered as a black box is used. The results show a good performance of the suggested approach. {\circledC}2008 AACC.",
author = "P{\'e}rez-Cruz, {J. Humberto} and Alexander Poznyak",
year = "2008",
month = "9",
day = "30",
doi = "10.1109/ACC.2008.4586817",
language = "American English",
pages = "2190--2195",
note = "Proceedings of the American Control Conference ; Conference date: 30-09-2008",

}

Pérez-Cruz, JH & Poznyak, A 2008, 'Neural control for power ascent of a TRIGA reactor' Paper presented at Proceedings of the American Control Conference, 30/09/08, pp. 2190-2195. https://doi.org/10.1109/ACC.2008.4586817

Neural control for power ascent of a TRIGA reactor. / Pérez-Cruz, J. Humberto; Poznyak, Alexander.

2008. 2190-2195 Paper presented at Proceedings of the American Control Conference, .

Research output: Contribution to conferencePaperResearch

TY - CONF

T1 - Neural control for power ascent of a TRIGA reactor

AU - Pérez-Cruz, J. Humberto

AU - Poznyak, Alexander

PY - 2008/9/30

Y1 - 2008/9/30

N2 - A basic control problem in a nuclear research reactor consists of increasing or decreasing the neutron power from a certain level R0 to a new desired level R1 and maintain the reactor stable at the new power level. For security reasons, this task must be performed in such a way that, during the power ascent, the instantaneous period of the reactor must always be greater than or equal to a lower limit value. To solve this problem, avoiding the difficulties associated with the physical modeling of the nuclear process, in this paper, we propose to use an indirect adaptive control scheme in which a single layer second order differential neural network achieves the on-line identification based only on three variables: the external reactivity, the fuel temperature, and the neutron power. The mathematical model provided by this identification process is employed to accomplish the control action in two stages. During the transient stage, the controller objective is to maintain the plant on a constant period. Once the desired power is reached, the control action is switched to a regulation stage. This identifier-controller is tested by simulation. Instead of the real plant, an eighth order physical model of a TRIGA reactor considered as a black box is used. The results show a good performance of the suggested approach. ©2008 AACC.

AB - A basic control problem in a nuclear research reactor consists of increasing or decreasing the neutron power from a certain level R0 to a new desired level R1 and maintain the reactor stable at the new power level. For security reasons, this task must be performed in such a way that, during the power ascent, the instantaneous period of the reactor must always be greater than or equal to a lower limit value. To solve this problem, avoiding the difficulties associated with the physical modeling of the nuclear process, in this paper, we propose to use an indirect adaptive control scheme in which a single layer second order differential neural network achieves the on-line identification based only on three variables: the external reactivity, the fuel temperature, and the neutron power. The mathematical model provided by this identification process is employed to accomplish the control action in two stages. During the transient stage, the controller objective is to maintain the plant on a constant period. Once the desired power is reached, the control action is switched to a regulation stage. This identifier-controller is tested by simulation. Instead of the real plant, an eighth order physical model of a TRIGA reactor considered as a black box is used. The results show a good performance of the suggested approach. ©2008 AACC.

U2 - 10.1109/ACC.2008.4586817

DO - 10.1109/ACC.2008.4586817

M3 - Paper

SP - 2190

EP - 2195

ER -

Pérez-Cruz JH, Poznyak A. Neural control for power ascent of a TRIGA reactor. 2008. Paper presented at Proceedings of the American Control Conference, . https://doi.org/10.1109/ACC.2008.4586817