Search for: All records

A Machine-Learning (ML) based detection scheme that automatically detects Alfvén Eigenmodes (AE) in a labelled DIII-D database is presented here. Controlling AEs is important for the success of planned burning plasma devices such as ITER, since resonant fast ions can drive AEs unstable and degrade the performance of the plasma or damage the first walls of the machine vessel. Artificial Intelligence could be useful for real-time detection and control of AEs in steady-state plasma scenarios by implementing ML-based models into control algorithms that drive actuators for mitigation of AE impacts. Thus, the objective is to compare differences in performance between using two different recurrent neural network systems (Reservoir Computing Network and Long Short Term Memory Network) and two different representations of the$\mathrm{C}{\mathrm{O}}_2$phase data (simple and crosspower spectrograms). All$\mathrm{C}{\mathrm{O}}_2$interferometer chords are used to train both models, but only one is processed during each training step. The results from the model and data comparison show higher performance for the RCN model (True Positive Rate = 90% and False Positive Rate = 14%), and that using simple magnitude spectrograms is sufficient to detect AEs. Also, the vertical$\mathrm{C}{\mathrm{O}}_2$interferometer chord passing near the center is better for ML-based detection of AEs.

 

Abstract Non-equilibrium plasmas derive their low temperature reactivity from producing and driving energetic electrons and active species under large electric fields. Therefore, the impact of reactants on the plasma properties including electron number density, electric field, and electron temperature is critical for applications such as plasma methane (CH 4 ) reforming. Due to experimental complexity, electron properties and the electric field are rarely measured together in the same discharge. In this work, we combine time-resolved Thomson scattering and electric field induced second harmonic generation to probe electron temperature, electron density, and electric field strength in a 60 Torr CH 4 /Ar nanosecond-pulsed dielectric barrier discharge while varying the CH 4 mole fraction from 0% to 8%. These measurements are compared to a 1D numerical model to benchmark its predictions and identify areas of uncertainty. Nonlinear coupling between CH 4 addition, electron temperature, electron density, and the electric field was directly observed. Contrary to previous measurements in He, the electron temperature increased with CH 4 mole fraction. This rise in electron temperature is identified as electron heating by residual electric fields that increased with larger CH 4 mole fraction. Moreover, the electron number density has been found to decrease rapidly with the increase of methane mole fraction. Comparison of these measurements with the model yielded better agreement at higher CH 4 mole fractions and with the usage of ab initio calculated Ar electron-impact cross-sections from the B-spline R-matrix database. Furthermore, the calculated plasma properties are shown to be sensitive to the residual surface charge implanted on the quartz dielectric surfaces. Without considering surface charge in the simulations, the calculated electric field profiles agreed well with the measurements, but the electron properties were underpredicted by more than a factor of three. Therefore, measurements of either the electric field or electron properties measurements alone are insufficient to fully validate modeling predictions. 

Abstract Alfvén eigenmodes (AEs) are an important and complex class of plasma dynamics commonly observed in tokamaks and other plasma devices. In this work, we manually labeled a small database of 26 discharges from the DIII-D tokamak in order to train simple neural-network-based models for classifying AEs. The models provide spatiotemporally local identification of four types of AEs by using an array of 40 electron cyclotron emission (ECE) signals as inputs. Despite the minimal dataset, this strategy performs well at spatiotemporally localized classification of AEs, indicating future opportunities for more sophisticated models and incorporation into real-time control strategies. The trained model is then used to generate spatiotemporally-resolved labels for each of the 40 ECE measurements on a much larger database of 1112 DIII-D discharges. This large set of precision labels can be used in future studies for advanced deep predictors and new physical insights. 

We report the development of a simple and sensitive two-beam hybrid femtosecond/picosecond pure rotational coherent anti-Stokes Raman scattering (fs/ps CARS) method to simultaneously measure the rotational and vibrational temperatures of diatomic molecules. Rotation–vibration non-equilibrium plays a key role in the chemistry and thermalization in low-temperature plasmas as well as thermal loading of hypersonic vehicles. This approach uses time-domain interferences between ground state and vibrationally excited N₂molecules to intentionally induce coherence beating that leads to apparent non-Boltzmann distributions in the pure rotational spectra. These distortions enable simultaneous inference of both the rotational and vibrational temperatures. Coherence beating effects were observed in single-shot fs/ps CARS measurements of a 75 Torr N₂DC glow discharge and were successfully modeled for rotational and vibrational temperature extraction. We show that this method can be more sensitive than a pure rotational fs/ps CARS approach using a spectrally narrow probe pulse. Lastly, we experimentally measured the beat frequencies via Fourier transform of the time-domain response and obtained excellent agreement with the model.

 

Abstract Modern tokamaks have achieved significant fusion production, but further progress towards steady-state operation has been stymied by a host of kinetic and MHD instabilities. Control and identification of these instabilities is often complicated, warranting the application of data-driven methods to complement and improve physical understanding. In particular, Alfvén eigenmodes are a class of ubiquitous mixed kinetic and MHD instabilities that are important to identify and control because they can lead to loss of confinement and potential damage to the walls of a plasma device. In the present work, we use reservoir computing networks to classify Alfvén eigenmodes in a large labeled database of DIII-D discharges, covering a broad range of operational parameter space. Despite the large parameter space, we show excellent classification and prediction performance, with an average hit rate of 91% and false alarm ratio of 7%, indicating promise for future implementation with additional diagnostic data and consolidation into a real-time control strategy. 

We present one-dimensional (1-D) imaging of rotation-vibration non-equilibrium measured by two-beam pure rotational hybrid femtosecond/picosecond coherent anti-Stokes Raman scattering (fs/ps CARS). Simultaneous measurements of the spatial distribution of molecular rotation-vibration non-equilibrium are critical for understanding molecular energy transfer in low temperature plasmas and hypersonic flows. However, non-equilibrium CARS thermometry until now was limited to point measurements. The red shift of rotational energy levels by vibrational excitation was used to determine the rotational and vibrational temperatures from 1-D images of the pure rotational spectrum. Vibrational temperatures up to 5500 K were detected in a${\mathrm{C}\mathrm{H}}_4/{\mathrm{N}}_2$nanosecond-pulsed pin-to-pin plasma within 2 mm near the cathode. This approach enables study of non-equilibrium systems with 40 µm spatial resolution.