Burn conditions stabilization with artificial neural networks of subignited thermonuclear reactors with scaling law uncertainties

Vitela, Javier E.; Martinell, Julio J.

doi:10.1088/0741-3335/43/2/302

Cited by 16 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The controller works for suppressing both thermal excursions and quenches, can operate at sub-ignition and ignition points (or points near the minimum power required for current drive), and can drive the system from one point to another during operation. Only those works that use non-model based control techniques, like neural networks [24,25], have also followed these guidelines. A zero-dimensional (volume-averaged) simulation study was performed to show the capabilities of the model-based controller and compare it with previous linear controllers.…”

Section: Prior Workmentioning

confidence: 99%

Nonlinear burn condition control in tokamaks using isotopic fuel tailoring

Boyer¹,

Schuster²

2015

Nucl. Fusion

View full text Add to dashboard Cite

One of the fundamental problems in tokamak fusion reactors is how to control the plasma density and temperature in order to regulate the amount of fusion power produced by the device. Control of these parameters will be critical to the success of burning plasma experiments like ITER. The most previous burn condition control efforts use either non-model based control designs or techniques based on models linearized around particular operating points. Such strategies limit the potential operational space and must be carefully retuned or redesigned to accommodate changes in operating points or plasma parameters. In this work, a nonlinear dynamic model of the spatial averages of energy and ion species densities is used to synthesize a nonlinear feedback controller for stabilizing the burn condition. The nonlinear model-based control strategy guarantees a much larger operational space than previous linear controllers. Because it is not designed around a particular operating point, the controller can be used to move from one burn condition to another. The proposed scheme first attempts to use regulation of the auxiliary heating power to reject temperature perturbations, then, if necessary, uses isotopic fuel tailoring as a way to reduce fusion heating during positive temperature perturbations. A global model of hydrogen recycling is incorporated into the model used for design and simulation, and the proposed control scheme is tested for a range of recycling model parameters. As we find the possibility of changing the isotopic mix can be limited for certain unfavorable recycling conditions, we also consider impurity injection as a backup method for controlling the system. A simple supervisory control strategy is proposed to switch between the primary and backup control schemes based on stability and performance criteria. A zero-dimensional simulation study is used to study the performance of the control scheme for several scenarios and model parameters. Finally, a one-dimensional simulation is done to test the robustness of the control scheme to spatially varying parameters.

show abstract

Section: Prior Workmentioning

confidence: 99%

Nonlinear burn condition control in tokamaks using isotopic fuel tailoring

Boyer¹,

Schuster²

2015

Nucl. Fusion

View full text Add to dashboard Cite

show abstract

“…The thermal (in-)stability of the operating point depends on the slope of the left hand side in equation ( 6) at the intersection point. If this slope exceeds 1, then not only does the natural numerical iteration scheme to solve equation ( 6) diverge, but the operating point is also thermally unstable (owing to the time derivative of the temperature term in equation ( 6)), so that active control (see, for example, [72]) is needed to keep the plasma at constant temperature.…”

Section: Uncertainty Propagation While Maximizing Qmentioning

confidence: 99%

On estimating the epistemic probability of realizing Q = Pfus/Paux larger than a specified lower bound in ITER

Kardaun¹

2002

Nucl. Fusion

View full text Add to dashboard Cite

A simplified analysis is given of the problem of estimating the `epistemic probability' for ITER to attain a power amplification factor Q larger than a certain lower bound. Attention is restricted to the parameters of ITER-FEAT and the 1998 ITER design. The probabilistic framework is briefly discussed and the distributional interval estimates for Q following from the interval estimates of the confinement time are derived (a) for ITER operation at constant fusion output power Pfus, and (b) for operation at a temperature that maximizes Q = Pfus/Paux. The second situation requires a radial integration of the flux surface averaged energy balance. Instead of a local transport model, a simple class of temperature profiles is used. The results are represented graphically. A generalization of the widely used fusion triple product plot against temperature is suggested. The analysis presupposes that at the reference operating point in situation (a), and in the range of operating temperatures projected to be achievable in situation (b), the conditions for reaching standard ELMy H mode in ITER are met. The practical conclusion of the article is that under this premise ITER-FEAT has a fairly large epistemic probability of obtaining plasma conditions with prevalent alpha particle heating.

show abstract

“…The controller works for suppressing both thermal excursions and quenches, can operate at sub-ignition and ignition points (or points near the minimum power required for current drive), and can drive the system from one point to another during operation. Only those works that use nonmodel based control techniques, like neural networks [22,23], have also followed these guidelines. A zero-dimensional (volume-averaged) simulation study was performed to show the capabilities of the model-based controller and compare it with previous linear controllers.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear control and online optimization of the burn condition in ITER via heating, isotopic fueling and impurity injection

Boyer¹,

Schuster²

2014

Plasma Phys. Control. Fusion

View full text Add to dashboard Cite

The ITER tokamak, the next experimental step toward the development of nuclear fusion reactors, will explore the burning plasma regime in which the plasma temperature is sustained mostly by fusion heating. Regulation of the fusion power through modulation of fueling and external heating sources, referred to as burn control, is one of the fundamental problems in burning plasma research. Active control will be essential for achieving and maintaining desired operating points, responding to changing power demands, and ensuring stable operation. Most existing burn control efforts use either non-model-based control techniques or designs based on linearized models. These approaches must be designed for particular operating points and break down for large perturbations. In this work, we utilize a spatially averaged (zero-dimensional) nonlinear model to synthesize a multi-variable nonlinear burn control strategy that can reject large perturbations and move between operating points. The controller uses all of the available actuation techniques in tandem to ensure good performance, even if one or more of the actuators saturate. Adaptive parameter estimation is used to improve the model parameter estimates used by the feedback controller in real-time and ensure asymptotic tracking of the desired operating point. In addition, we propose the use of a model-based online optimization algorithm to drive the system to a state that minimizes a given cost function, while respecting input and state constraints. A zero-dimensional simulation study is presented to show the performance of the adaptive control scheme and the optimization scheme with a cost function weighting the fusion power and temperature tracking errors.

show abstract

Burn conditions stabilization with artificial neural networks of subignited thermonuclear reactors with scaling law uncertainties

Cited by 16 publications

References 23 publications

Nonlinear burn condition control in tokamaks using isotopic fuel tailoring

Nonlinear burn condition control in tokamaks using isotopic fuel tailoring

On estimating the epistemic probability of realizing Q = Pfus/Paux larger than a specified lower bound in ITER

Nonlinear control and online optimization of the burn condition in ITER via heating, isotopic fueling and impurity injection

Contact Info

Product

Resources

About