Artificial Neural Network (ANN) modeling of the pulsed heat load during ITER CS magnet operation

Savoldi, Laura; Bonifetto, Roberto; Carli, Stefano; Froio, Antonio; Foussat, A.; Zanino, Roberto

doi:10.1016/j.cryogenics.2014.03.003

Cited by 18 publications

(9 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial intelligence methods are powerful tools to accurately model certain data behaviour, which have been recently involved in solving applied superconductivity problems [50]. For instance, artificial neural network (ANN) has been utilized in [51][52][53][54] to estimate the calculation of AC losses in specific SC system and aid the design of SC facilities. As an alternative, we have explored the feasibility of constructing a statistical model, which can give accurate V oc values according to the given controllable parameters within seconds, and hence provides primary guidance for the design of such devices in practical applications.…”

Section: Regression Analysismentioning

confidence: 99%

Sensitivity analysis and machine learning modelling for the output characteristics of rotary HTS flux pumps

Wen

Zhang

Mueller

2021

Supercond. Sci. Technol.

View full text Add to dashboard Cite

High temperature superconducting (HTS) rotatory flux pump, or so called HTS dynamo, can output none-zero time-averaged DC voltage and charge the rest of the circuit if a closed loop has been formed. This type of flux pump is often employed together with HTS coils, where the HTS coils can potentially work in persistent current mode, and act like electromagnets with considerable magnetic field, having wide range of applications in industry. The output characteristics of HTS rotary flux pumps have been extensively explored through experiments and finite element method (FEM) simulations, yet the work on constructing statistical models as an alternative approach to capture key characteristics has not been studied and published. A 2D FEM program has been used to model the HTS rotatory flux pumps and evaluate the effects of different factors upon the output voltage through parameter sweeping and analysis of variance. Typical design considerations, including operation frequency, air gap, HTS tape width and remanent flux density have been investigated, in particular the bilateral effect of HTS tape width has been explained by looking at the averaged integration of the electric field over the tape. Based on the data obtained from various simulations, regression analysis has been conducted through a collection of machine learning methods and demonstrated that the output voltage of a rotary flux pump can be obtained promptly with satisfactory accuracy via Gaussian process regression, aiming to provide a novel approach for future research and powerful design tool for industrial applications using HTS rotary flux pump devices.

show abstract

Section: Regression Analysismentioning

confidence: 99%

Sensitivity analysis and machine learning modelling for the output characteristics of rotary HTS flux pumps

Wen

Zhang

Mueller

2021

Supercond. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Supercritical Helium (SHe) is kept in motion in the SMCCC to cool down each CSM [68]. Figure 1 shows a simplified scheme of the SMCCC circuit with its main components.…”

Section: The Superconducting Magnet Cryogenic Cooling Circuit (Smccc)mentioning

confidence: 99%

“…The closed cooling circuit is simulated for a mission time t miss = 3600 s (at the beginning of which the current evolution foreseen for the ITER 15 MA plasma scenario is followed) with the 4C code that includes [35]: (i) a 1-D thermal-hydraulic model for each channel of the CSM that is thermally coupled with the others through a 2 1/2 -D model accounting for heat diffusion phenomena in each radial section of the CSM; (ii) a 1-D compressible fluid model for pipes and HXs; (iii) a 0-D model for the mass and the energy balance in relevant points of the cooling loop (such as valves, QT, CP, etc.). The choice of the mission time t miss = 3600 s is dictated by the length and shape of one single pulse of current in the ITER CSM, which can be divided into Q = 5 phases: Further physical details are not reported here for brevity, since they go far beyond the methodological scope of the present work: the interested reader is referred to [48,68].…”

mentioning

confidence: 99%

Metamodeling and On-Line Clustering for Loss-of-Flow Accident Precursors Identification in a Superconducting Magnet Cryogenic Cooling Circuit

et al. 2021

View full text Add to dashboard Cite

In the International Thermonuclear Experimental Reactor, plasma is magnetically confined with Superconductive Magnets (SMs) that must be maintained at the cryogenic temperature of 4.5 K by one or more Superconducting Magnet Cryogenic Cooling Circuits (SMCCC). To guarantee cooling, Loss-Of-Flow Accidents (LOFAs) in the SMCCC are to be avoided. In this work, we develop a three-step methodology for the prompt detection of LOFA precursors (i.e., those combinations of component failures causing a LOFA). First, we randomly generate accident scenarios by Monte Carlo sampling of the failures of typical SMCCC components and simulate the corresponding transient system response by a deterministic thermal-hydraulic code. In this phase, we also employ quick-running Proper Orthogonal Decomposition (POD)-based Kriging metamodels, adaptively trained to reproduce the output of the long-running code, to decrease the computational time. Second, we group the generated scenarios by a Spectral Clustering (SC) employing the Fuzzy C-Means (FCM), in order to identify the main patterns of system evolution towards abnormal states (e.g., a LOFA). Third, we develop an On-line Supervised Spectral Clustering (OSSC) technique to associate time-varying parameters measured during plant functioning to one of the prototypical groups obtained, which may highlight the related LOFA precursors (in terms of SMCCC components failures). We apply the proposed technique to the simplified model of a cryogenic cooling circuit of a single module of the ITER Central Solenoid Magnet (CSM). The framework developed promptly detects 95% of LOFA events and around 80% of the related precursors.

show abstract

“…In [35], neural networks are used to accelerate thermal-hydraulic modeling and enables faster assessment for the dynamic thermal-hydraulic system. Richard et al [34] apply a neural network to ITER magnets to predict the occurrence of the interruption, which can be used to adjust the reaction to continue generating power and avoid ITER damage. Mathuriya et al [36] build a CNN model to determine the physical model that describes our universe.…”

Section: Related Workmentioning

confidence: 99%

“…Neural networks (especially deep neural networks (DNNs)) have been employed in HPC applications recently. In [34], a neural network is used to generate input data for modulating the simulation process. Wigley et al [7] propose a neural network-based online optimization process for the Bose-Einstein condensates (BEC).…”

Section: Related Workmentioning

confidence: 99%

Adaptive neural network-based approximation to accelerate eulerian fluid simulation

Dong

Liu

Xie

et al. 2019

Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

The Eulerian fluid simulation is an important HPC application. The neural network has been applied to accelerate it. The current methods that accelerate the fluid simulation with neural networks lack flexibility and generalization. In this paper, we tackle the above limitation and aim to enhance the applicability of neural networks in the Eulerian fluid simulation. We introduce Smart-fluidnet, a framework that automates model generation and application. Given an existing neural network as input, Smart-fluidnet generates multiple neural networks before the simulation to meet the execution time and simulation quality requirement. During the simulation, Smartfluidnet dynamically switches the neural networks to make best efforts to reach the user's requirement on simulation quality. Evaluating with 20,480 input problems, we show that Smart-fluidnet achieves 1.46x and 590x speedup comparing with a state-of-the-art neural network model and the original fluid simulation respectively on an NVIDIA Titan X Pascal GPU, while providing better simulation quality than the state-of-the-art model. CCS CONCEPTS • Computing methodologies → Parallel algorithms; Neural networks; Model development and analysis.

show abstract

Artificial Neural Network (ANN) modeling of the pulsed heat load during ITER CS magnet operation

Cited by 18 publications

References 13 publications

Sensitivity analysis and machine learning modelling for the output characteristics of rotary HTS flux pumps

Sensitivity analysis and machine learning modelling for the output characteristics of rotary HTS flux pumps

Metamodeling and On-Line Clustering for Loss-of-Flow Accident Precursors Identification in a Superconducting Magnet Cryogenic Cooling Circuit

Adaptive neural network-based approximation to accelerate eulerian fluid simulation

Contact Info

Product

Resources

About