Machine learning in analytical spectroscopy for nuclear diagnostics [Invited]

Ashwin P. Rao; Phillip R. Jenkins; Ryan E. Pinson; John D. Auxier II; Michael B. Shattan; Anil K. Patnaik

doi:10.1364/AO.482533

Applied Optics
Vol. 62,
Issue 6,
pp. A83-A109
(2023)
•https://doi.org/10.1364/AO.482533

Machine learning in analytical spectroscopy for nuclear diagnostics [Invited]

Ashwin P. Rao, Phillip R. Jenkins, Ryan E. Pinson, John D. Auxier II, Michael B. Shattan, and Anil K. Patnaik

Get PDF
Email
Share
Get Citation
Copy Citation Text
Ashwin P. Rao, Phillip R. Jenkins, Ryan E. Pinson, John D. Auxier II, Michael B. Shattan, and Anil K. Patnaik, "Machine learning in analytical spectroscopy for nuclear diagnostics [Invited]," Appl. Opt. 62, A83-A109 (2023)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Check for updates

Related Topics
Optics & Photonics Topics
?

The topics in this list come from the Optics and Photonics Topics applied to this article.

About this Article
History
- Original Manuscript: December 6, 2022
- Revised Manuscript: January 23, 2023
- Manuscript Accepted: January 24, 2023
- Published: February 14, 2023

Accepted Manuscript
Download Accepted Manuscript
Access to this article is made available via and is subject to Optica Publishing Group Terms of Use.

Abstract

Analytical spectroscopy methods have shown many possible uses for nuclear material diagnostics and measurements in recent studies. In particular, the application potential for various atomic spectroscopy techniques is uniquely diverse and generates interest across a wide range of nuclear science areas. Over the last decade, techniques such as laser-induced breakdown spectroscopy, Raman spectroscopy, and x-ray fluorescence spectroscopy have yielded considerable improvements in the diagnostic analysis of nuclear materials, especially with machine learning implementations. These techniques have been applied for analytical solutions to problems concerning nuclear forensics, nuclear fuel manufacturing, nuclear fuel quality control, and general diagnostic analysis of nuclear materials. The data yielded from atomic spectroscopy methods provide innovative solutions to problems surrounding the characterization of nuclear materials, particularly for compounds with complex chemistry. Implementing these optical spectroscopy techniques can provide comprehensive new insights into the chemical analysis of nuclear materials. In particular, recent advances coupling machine learning methods to the processing of atomic emission spectra have yielded novel, robust solutions for nuclear material characterization. This review paper will provide a summation of several of these recent advances and will discuss key experimental studies that have advanced the use of analytical atomic spectroscopy techniques as active tools for nuclear diagnostic measurements.

Full Article | PDF Article

More Like This

LIBS and Raman spectroscopy in tandem with machine learning for interrogating weatherization of lithium hydride

Ryan E. Pinson, Andrew V. Giminaro, Christina L. Dugan, Phillip R. Jenkins, and Anil K. Patnaik
Appl. Opt. 62(6) A118-A126 (2023)

Development of advanced machine learning models for analysis of plutonium surrogate optical emission spectra

Ashwin P. Rao, Phillip R. Jenkins, John D. Auxier, Michael B. Shattan, and Anil K. Patnaik
Appl. Opt. 61(7) D30-D38 (2022)

Recent progress in high-speed laser diagnostics for hypersonic flows [Invited]

Naibo Jiang, Paul S. Hsu, Mark Gragston, and Sukesh Roy
Appl. Opt. 62(6) A59-A75 (2023)

Previous Article Next Article

Data availability

The data in these investigations are not publicly available but may be accessible to individuals on a case-by-case basis. Requests should be referred to the corresponding author for proper routing.

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (37)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (15)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (12)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

	PCR	PLSR
RMSE	0.716	0.216
$R^{2}$	0.603	0.964

$U$ line (nm)	$R^{2}$	LoD (wt%)
409.0	0.990	0.38
502.7	0.993	0.88
591.5	0.953	1.45
682.7	0.986	0.54

Method	Fit	$R^{2}$	RMSE
LIBS	Univariate	0.66	13.5
	PCR	0.65	11.1
	PLSR	0.89	6.66
Raman	Univariate	0.93	5.39
	PCR	0.74	11.1
	PLSR	0.73	10.0
Fusion	PCR	0.87	6.67
Fusion	PLSR	0.98	2.47

Laser	Nd:YAG
Wavelength	1064 nm
Pulse width	1 ns
Pulse energy	5–6 mJ
Focal length	1.5 cm
Spot size	$50 µ m$
Dimensions	$8.25 \times 11.5 \times 4.5 i n$
Weight	4 lbs
Bandwidth	190–950 nm
Resolution	0.1 nm FWHM

Model	$R^{2}$	RMSE	LoD
PCR	0.887	1.388%	1.67%
PLSR	0.967	0.749%	1.15%

Model	$R^{2}$	RMSEP	LoD
PCR	0.8871	1.388%	1.669 %
PLSR	0.967	0.749%	1.155%
Bagged trees	0.974	0.675%	0.279%
Boosted trees	0.964	0.272%	2.05%
ANN	0.936	1.059%	1.086%

Hardware	Model	Parameters
Laser	Quantel Everbright 250	15 Hz rep rate; 10 ns pulse width
Spectrometer	Catalina Scientific EMU-120/65	$30 \times 120 µ m$ slit width; 25 mm AS;
Camera	Andor USB iStar	$1024 \times 1024$ pixel CCD
Delay generator	Berkeley Nucleonics 577 DDG	–

Model	Hyperparameters	Range	Tuned Value
Tree	Min. leaf size	1–144	20
	Max mum. splits	1–100	61
Bag	Min. leaf size	1–144	10
	Num. learning cycles	10–500	495
Boost	Max num. splits	1–100	34
	Num. learning cycles	10–500	180
	Learning rate	0.001–1	0.095
ET	Min. leaf size	1–144	20
	Max num. splits	1–100	5
	Num. learning cycles	10–500	300
RF	Min. leaf size	1–144	20
	Max num. splits	1–100	10
	Num. learning cycles	10–500	300
SVR	Kernel function	Linear; Gaussian; polynomial	Linear
	Slack ( $ξ$ )	0.001–1000	200.2
	Bandwidth ( $h$ )	0.001–1000	454.7
	Error ( $ϵ$ )	1.48e-3–148	0.362
GKR	Bandwidth ( $h$ )	0.001–1000	51.14
	Regularization ( $λ$ )	4.99e-7–0.499	1.72e-5
	Error ( $ϵ$ )	1.48e-3–148	1.51e-3
ANN	Layer size	Narrow; medium; wide; bilayer; trilayer	Trilayer
	Number of neurons	2–50	[20;10;10]
	Activation function	ReLU; sigmoid, tanh	Sigmoid
	Iteration limit	1e2-1e4	1000
	Regularization ( $λ$ )	4.99e-7–0.499	4.96e-5

Model	Tree	Bag	Boost	ET	RF
RMSEP	0.475%	0.394%	0.422%	0.394%	0.391%
LoD	0.366%	0.025%	0.256%	0.006%	0.018%

Site	Debris Type	Yield	HOB	# Samples
1	Particle	500 t	$- 1 m$	11
2	Deposition	500 t	$- 30 m$	10
3	Particle	22 t	$+ 1 m$	9
4	Deposition	30 kt	$+ 1 m$	8
5	Particle	18 t	$- 110 m$	10
6	Particle/deposition	44 kt	$+ 210 m$	18

Site	Na	Mg	Al	Si	K	Ca	Ti	Cr	Mn	Fe	Ba	Li
1	3.01	0.18	27.7	11.3	13. 5	3.16	0.09	0.14	0.11	1.68	0.04	0.01
2	4.85	0.45	17.8	6.08	15.7	7.88	0.09	0.08	0.12	3.4	0.09	0.02
3	8.98	4.25	13.61	32.9	25.3	3.81	1.32	2.50	0.09	7.92	0.49	0.01
4	2.30	0.42	19.3	41.3	10.7	2.14	0.19	0.34	0.03	0.97	0.16	0.004
5	2.23	0.35	9.90	17.1	9.62	4.35	0.04	0.03	0.12	9.49	0.05	0.02
6	2.90	0.61	15.9	42.5	18.0	2.64	0.43	1.14	0.15	5.70	0.29	0.006
Trinity	2.68	1.13	15.0	66	4.32	6.87	0.42	0.01	0.08	2.27	0.1	–

Hyperparameter	Search Values	Tuned Value
# of trees	100–1000	1000
# Predictors per split	10–1000	1000
Minimum # of leaves	1–100	1
# of splits	10–1000	1000

		Humidity		Temperature
Data Set	Method	$R^{2}$	RMSE	$R^{2}$	RMSE
LIBS	PLSR	.84	11.48	.66	8.42
	Random forest	.88	8.10	.73	3.47
Raman	PLSR	.70	15.2	.43	11.9
	Random forest	98	3.90	.74	4.65
Fusion	PLSR	.91	8.22	.65	8.04
	Random forest	.98	3.79	.85	3.37

Ashwin P. Rao	https://orcid.org/0000-0003-1931-2568
Phillip R. Jenkins	https://orcid.org/0000-0002-2425-9151
John D. Auxier II	https://orcid.org/0000-0001-6234-6451
Michael B. Shattan	https://orcid.org/0000-0003-2476-1253
Anil K. Patnaik	https://orcid.org/0000-0001-7686-4305

Model	SVR	GKR	ANN
RMSEP	0.611%	0.329%	0.399%
LoD	0.098%	0.015%	0.017%

Peak	MAPE	LoD
$K_{α}$	9.8 %	0.002%
$K_{β}$	8.3%	0.008%

Model	SVR	GKR	ANN
RMSEP	0.611%	0.329%	0.399%
LoD	0.098%	0.015%	0.017%

Abstract

Data availability

Cited By

Figures (37)

Tables (15)

Equations (12)

Applied Optics