A methodology is proposed, which addresses the caveat that line-of-sight emission spectroscopy presents in it cannot provide spatially resolved temperature measurements non-homogeneous fields. The aim of this research to explore use data-driven models measuring distributions a manner using data. Two categories methods are analyzed: (i) Feature engineering and classical machine learning algorithms, (ii) end-to-end convolutional neural networks (CNN). In total, combinations fifteen feature...

Machine learning (ML) techniques are popular in many parameter estimation tasks; however, they face challenges the real-world deployment due to lack of robustness errors. ML estimators not able ascertain performance presence noise, variations data distribution, and anomalies test samples. This work proposes a novel framework, surrogate-based physical error correction (SPEC), which addresses unmet need for measurement reliability self-correction under process uncertainty, by bringing together...

Forward modelling of seismic wavefields is a cornerstone geophysical studies, aiding in subsurface characterization and exploration. In this study, we introduce Physics-Enhanced Deep Fourier-Attention Network (PE-DFAN) to simulate the forward process from physical property differences wavefields, addressing limitations conventional neural networks capturing complex wavefield patterns. Conventional often struggle model intricate spatial correlations inherent wave propagation, resulting...

We present a novel approach to gravity forward modeling using conditional neural operators that establishes generative model from the basin models and hyperparameters (reference basement depth, etc.) anomaly. Our methodology introduces an innovative adaptive embedding mechanism where scalar are first embedded into 32-dimensional space then adaptively expanded match dimensions of depth model, enabling effective fusion with data. Subsequently, Fourier Convolution Layers employed transform...

Laser absorption spectroscopy (LAS) quantification is a popular tool used in measuring temperature and concentration of gases. It has low error tolerance, whereas current ML-based solutions cannot guarantee their measure reliability. In this work, we propose new framework, SPEC, to address issue. addition the conventional ML estimator-based estimation mode, SPEC also includes Physics-driven Anomaly Detection module (PAD) assess estimation. And Correction mode designed correct unreliable The...

Electrical impedance tomography (EIT) is a non-invasive imaging technique, capable of reconstructing images the electrical conductivity tissues and materials. It popular in diverse application areas, from medical to industrial process monitoring tactile sensing, due its low cost, real-time capabilities non-ionizing nature. EIT visualizes distribution within body by measuring boundary voltages, given current injection. However, image reconstruction ill-posed mismatch between under-sampled...

When deploying machine learning estimators in science and engineering (SAE) domains, it is critical to avoid failed estimations that can have disastrous consequences, e.g., aero engine design. This work focuses on detecting correcting state before adopting them SAE inverse problems, by utilizing simulations performance metrics guided physical laws. We suggest flag a estimation when its model error exceeds feasible threshold, propose novel approach, GEESE, correct through optimization, aiming...

In this paper, a high speed Turbojet-Scramjet combined engine is constructed, and the performance calculation completed. The mission of long range strike set as target aircraft's main mission. A corresponding flight profile was found. Aircraft/power integrated design model developed, which used to evaluate aircraft finishing results show that can work in Ma 0-6 0-25km airspace. assembled complete high-altitude high-speed long-range

Physics-based inverse modeling techniques are typically restricted to particular research fields, whereas popular machine-learning-based ones too data-dependent guarantee the physical compatibility of solution. In this paper, Self-Validated Physics-Embedding Network (SVPEN), a general neural network framework for is proposed. As its name suggests, embedded forward model ensures that any solution successfully passes validation physically reasonable. SVPEN operates in two modes: (a) function...

The current success of machine learning on image-based combustion monitoring is based massive data, which costly even impossible for industrial applications. To address this conflict, we introduce few-shot in order to achieve and classification the first time. Two algorithms, Siamese Network coupled with k Nearest Neighbors (SN-kNN) Prototypical (PN), were tested. Rather than utilizing solely visible images as discussed previous studies, also used Infrared (IR) images. We analyzed training...

A network-based optimization approach, EEE, is proposed for the purpose of providing validation-viable state estimations to remediate failure pretrained models. To improve efficiency and convergence, most important metrics in context this research, we follow a three-faceted approach based on error from validation process. Firstly, information content by designing module acquire high-dimensional information. Next, reduce uncertainty transfer employing an ensemble estimators, which only learn...

The effect of spatial nonuniformity the temperature distribution was examined on capability machine-learning algorithms to provide accurate prediction based Laser Absorption Spectroscopy. First, sixteen machine learning models were trained as surrogate conventional physical methods measure from uniform distributions (uniform-profile spectra). best three them, Gaussian Process Regression (GPR), VGG13, and Boosted Random Forest (BRF) shown work excellently profiles but their performance...

A methodology is proposed, which addresses the caveat that line-of-sight emission spectroscopy presents in it cannot provide spatially resolved temperature measurements nonhomogeneous fields. The aim of this research to explore use data-driven models measuring distributions a manner using data. Two categories methods are analyzed: (i) Feature engineering and classical machine learning algorithms, (ii) end-to-end convolutional neural networks (CNN). In total, combinations fifteen feature...

Abstract Some studies have qualitatively proved that nonuniform profiles along the light path in turbulent flows can cause temperature measurement inaccuracies Laser Absorption spectroscopy (LAS), based on analysis of Beer-Lambert’s law. In this work, we attempt to further analyze nonuniformity effect quantitatively from viewpoint data analysis. Ten thousand synthetic CO2 absorption spectra are respectively generated uniform and five discrete sections profiles. Sixteen machine learning/deep...

A methodology is proposed, which addresses the caveat that line-of-sight emission spectroscopy presents in it cannot provide spatially resolved temperature measurements nonhomogeneous fields. The aim of this research to explore use data-driven models measuring distributions a manner using data. Two categories methods are analyzed: (i) Feature engineering and classical machine learning algorithms, (ii) end-to-end convolutional neural networks (CNN). In total, combinations fifteen feature...