Deep learning for solving the Boltzmann equation of electrons in weakly ionized plasma

Kawaguchi, Satoru; Takahashi, Kazuhiro; Ohkama, Hiroshi; Satoh, Kohki

doi:10.1088/1361-6595/ab6074

Cited by 20 publications

(24 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…33,34 Automatic differentiation is again exploited to obtain desired partial derivatives with respect to the input variables. 35 Early work in LTP has been proposed by Kawaguchi et al 50 who have considered a PINN solution to the stationary Boltzmann equation subject to ionization and electron attachment processes. An MLP has been trained to predict the electron velocity distribution function (EVDF) following an exponential ansatz.…”

Section: Physics-informed Machine Learningmentioning

confidence: 99%

Review: Machine learning for advancing low-temperature plasma modeling and simulation

Trieschmann,

Vialetto,

Gergs

2023

J. Micro/Nanopattern. Mats. Metro.

View full text Add to dashboard Cite

Machine learning has had an enormous impact in many scientific disciplines. It has also attracted significant interest in the field of low-temperature plasma (LTP) modeling and simulation in past years. Its application should be carefully assessed in general, but many aspects of plasma modeling and simulation have benefited substantially from recent developments within the field of machine learning and datadriven modeling. In this survey, we approach two main objectives: (a) we review the state-of-the-art, focusing on approaches to LTP modeling and simulation. By dividing our survey into plasma physics, plasma chemistry, plasma-surface interactions, and plasma process control, we aim to extensively discuss relevant examples from literature. (b) We provide a perspective of potential advances to plasma science and technology. We specifically elaborate on advances possibly enabled by adaptation from other scientific disciplines. We argue that not only the known unknowns but also unknown unknowns may be discovered due to the inherent propensity of data-driven methods to spotlight hidden patterns in data.

show abstract

Section: Physics-informed Machine Learningmentioning

confidence: 99%

Review: Machine learning for advancing low-temperature plasma modeling and simulation

Trieschmann,

Vialetto,

Gergs

2023

J. Micro/Nanopattern. Mats. Metro.

View full text Add to dashboard Cite

show abstract

“…where B is the magnetic field. Finally, the electron density under the magnetic field can be calculated by Equation ( 9) in the experiment, the results measured by the probe will also be corrected by Equation (9) to train the machine learning model.…”

Section: Correction Of the Probementioning

confidence: 99%

“…Considering that under the influence of the magnetic field, the accuracy and parameter range of the probe diagnostic are greatly limited, and machine learning technology can break through this limitation by deeply mining information and summarizing physical laws. Recently, there have been some studies using machine learning to improve probe diagnostic methods, [ 9 ] and some studies have demonstrated the feasibility of applying machine learning technology in principle. [ 10 ] In addition, the machine learning algorithm is applied to plasma parameter diagnosis to improve the precision and measurement range of probe diagnostic.…”

Section: Introductionmentioning

confidence: 99%

Feasibility analysis of machine learning applied to magnetized plasma diagnosis

Guan

Zhang

et al. 2022

Contributions to Plasma Physics

View full text Add to dashboard Cite

In the paper, machine learning is demonstrated for the diagnostic enhancement of magnetized plasma probes. In the plasma experiment, the magnetic field greatly affects the properties of plasma and the performance of the probe, which limits the range of measurement parameters and reduces the accuracy of probe diagnostic results. Existing probe correction methods based on improved theory and mechanics cannot completely eliminate the influence of magnetic field and often introduce additional errors. In this paper, a novel machine learning method is proposed to improve magnetized plasma probe diagnostic based on existing methods and traditional probe correction theory. The pressure of the plasma, the total voltage of the circuit, and the magnetic induction intensity are used as input parameters, and the electron density obtained from the probe diagnostics are used as output parameters. The paper presents experiments to analyse the original probe results and the probe results revised by magnetic field probe theory through the machine learning algorithm and compare them with the results of theoretical simulations. The experimental results prove that the machine learning model based on revised data has better learning efficiency and prediction results, which can expand the application scope of traditional probe correction theory and predict results closer to the theoretical value.

show abstract

“…Kawaguchi et al. [40] used an artificial feed‐forward neural network to solve Boltzmann's equation for an electron velocity distribution function. Mesbah et al.…”

Section: Introductionmentioning

confidence: 99%

Modulating sparks in a pulse train for repetitive and energy efficient plasma generation

Yin

Zhu

2023

High Voltage

View full text Add to dashboard Cite

Spark is a widely studied plasma source for active species production; however, it experiences unstable transitions (e.g. to a thermal arc) at high frequencies or long pulse durations. In this study, the sparks generated in a pulse train were studied and modulated based on a physics‐corrected deep learning method. Our results show that a highly repeatable and stable spark plasma source can be achieved by automatically adjusting the voltage amplitude according to the discharge frequency in a high‐frequency pulse train within the time scale of the fluid response. The influences of the electron number density increasing mode and modulated driven voltage profiles on the energy efficiencies were also studied.

show abstract

Deep learning for solving the Boltzmann equation of electrons in weakly ionized plasma

Cited by 20 publications

References 27 publications

Review: Machine learning for advancing low-temperature plasma modeling and simulation

Review: Machine learning for advancing low-temperature plasma modeling and simulation

Feasibility analysis of machine learning applied to magnetized plasma diagnosis

Modulating sparks in a pulse train for repetitive and energy efficient plasma generation

Contact Info

Product

Resources

About