Calibration of spectra in presence of non-stationary background using unsupervised physics-informed deep learning

Puleio, Alessandro; Rossi, Riccardo; Gaudio, P

doi:10.1038/s41598-023-29371-9

Cited by 4 publications

(1 citation statement)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This process establishes theoretical constraints and biases to supplement measurement data. PINNs can be applied to both supervised and unsupervised learning tasks 7 – 9 , as well as to forward and inverse problems 10 . PINNs training process requires substantially less data than for most deep learning methods because PINN performance is not directly related to the volume of training data.…”

Section: Introductionmentioning

confidence: 99%

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Prantikos,

Chatzidakis,

Tsoukalas

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Nuclear reactor safety and efficiency can be enhanced through the development of accurate and fast methods for prediction of reactor transient (RT) states. Physics informed neural networks (PINNs) leverage deep learning methods to provide an alternative approach to RT modeling. Applications of PINNs in monitoring of RTs for operator support requires near real-time model performance. However, as with all machine learning models, development of a PINN involves time-consuming model training. Here, we show that a transfer learning (TL-PINN) approach achieves significant performance gain, as measured by reduction of the number of iterations for model training. Using point kinetic equations (PKEs) model with six neutron precursor groups, constructed with experimental parameters of the Purdue University Reactor One (PUR-1) research reactor, we generated different RTs with experimentally relevant range of variables. The RTs were characterized using Hausdorff and Fréchet distance. We have demonstrated that pre-training TL-PINN on one RT results in up to two orders of magnitude acceleration in prediction of a different RT. The mean error for conventional PINN and TL-PINN models prediction of neutron densities is smaller than 1%. We have developed a correlation between TL-PINN performance acceleration and similarity measure of RTs, which can be used as a guide for application of TL-PINNs.

show abstract

Section: Introductionmentioning

confidence: 99%

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Prantikos,

Chatzidakis,

Tsoukalas

et al. 2023

Sci Rep

View full text Add to dashboard Cite

show abstract

Multiparameter optical fiber sensing for energy infrastructure through nanoscale light–matter interactions: From hardware to software, science to commercial opportunities

Su,

Ohodnicki,

Wuenschell

et al. 2024

APL Photonics

View full text Add to dashboard Cite

Monitoring of energy infrastructure through robust yet economical sensing platforms is becoming an area of increased importance, with ubiquitous applications including the electrical grid, natural gas and oil transportation pipelines, H2 infrastructure (storage and transportation), carbon storage, power generation, and subsurface environments. Plasmonic and functional nanomaterial enabled fiber optic sensors show excellent promise for a wide range of sensing applications due to their versatility to be engineered for specific analytes of interest while retaining inherent advantages of the optical fiber sensor platform. Through the design of novel sensing layers, the optical transduction mechanism and wavelength dependence can also be tailored for ease of integration with low-cost interrogation systems enabling an inexpensive yet highly functional optical fiber sensing platform. In addition, recent advances in artificial intelligence and machine learning theoretical methods have been leveraged to simultaneously extract multiple parameters through multi-wavelength interrogation such that unique wavelengths can also serve as unique sensing elements, analogous to electronic nose sensor technologies. The concept of an optical fiber based “photonic nose” via multiple interrogation wavelengths and/or sensor nodes offers a compelling platform technology to realize multiparameter speciation of chemical analytes within complex gas mixtures. In this Perspective, we further generalize the notion of multiparameter sensing through the novel “photonic nervous system” concept based upon low-cost, functionalized optical fiber sensor probes monitoring a variety of distinct analyte classes (physical, chemical, electromagnetic, etc.) simultaneously to provide broad situational awareness via integrated sensors.

show abstract

Physics-Informed Neural Networks for Modeling Incompressible Laminar Flows with Mixed-Variable Formulation

Li,

Pan,

Chen

et al. 2024

Proceedings of the 2024 3rd International Conference on Frontiers of Artificial Intelligence and Machine Learning

View full text Add to dashboard Cite

Calibration of spectra in presence of non-stationary background using unsupervised physics-informed deep learning

Cited by 4 publications

References 63 publications

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Multiparameter optical fiber sensing for energy infrastructure through nanoscale light–matter interactions: From hardware to software, science to commercial opportunities

Physics-Informed Neural Networks for Modeling Incompressible Laminar Flows with Mixed-Variable Formulation

Contact Info

Product

Resources

About