Research on a rapid evaluation method for laser atmospheric propagation based on machine learning

Zhifu HUANG; Ying ZHANG; Nan LI; Xiaoxing FENG; Zhiqiang WANG; Chunhong QIAO; Chengyu FAN; Yingjian WANG

doi:10.3788/IRLA20250054

Infrared and Laser Engineering, Volume. 54, Issue 7, 20250054(2025)

Research on a rapid evaluation method for laser atmospheric propagation based on machine learning

Zhifu HUANG^1,2, Ying ZHANG^1,2, Nan LI^1、*, Xiaoxing FENG¹, Zhiqiang WANG^3,4, Chunhong QIAO¹, Chengyu FAN^3,4, and Yingjian WANG^3,4

Author Affiliations

¹Anhui Institute of Optics and Fine Mechanics, HFIPS, Chinese Academy of Sciences, State Key Laboratory of Laser Interaction with Matter, Hefei 230031, China

²Science Island Branch of Graduate School, University of Science and Technology of China, Hefei 230026, China

³Key Laboratory of Atmospheric Optics, Anhui Institute of Optics and Fine Mechanics, HFIPS, Chinese Academy of Sciences, Hefei 230031, China

⁴National University of Defense Technology, Nanhu Laser Laboratory, Changsha 410073, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. Lasso regression model construction steps

Download full size

View in Article

Fig. 2. MSE heat map of polynomial degrees and regularization intensity. (a) ER63.2%; (b) ER83.9%

Download full size

View in Article

Fig. 3. 2D comparison of the model evaluation results with the HELP-4D simulation results (ER63.2). (a) Training set; (b) Test set

Download full size

View in Article

Fig. 4. 2D comparison of the model evaluation results with the HELP-4D simulation results (ER83.9%). (a) Training set; (b) Test set

Download full size

View in Article

Fig. 5. 3D comparison of the model evaluation results with the HELP-4D simulation results (63.2% extension multiple). (a) Training set; (b) Test set

Download full size

View in Article

Fig. 6. 3D comparison of the model evaluation results with the HELP-4D simulation results (83.9% extension multiple). (a) Training set; (b) Test set

Download full size

View in Article

Fig. 7. Beam quality factor evaluation relative error distribution plot. (a) ER63.2%; (b) EM63.2%; (c) ER83.9%; (d) EM83.9%

Download full size

View in Article

Table 1. Data structure at different levels of the Lasso regression model

View table

View in Article

Table 1. Data structure at different levels of the Lasso regression model

Data level	Input dimension	Output dimension	Data transformation operations
Original input layer	10	10	Data standardization
Polynomial expansion layer	10	3003	High-dimensional nonlinear features
L1 regularization screening layer	3003	11	Feature screening
Output layer	11	1	Linear combination

Table 2. Data range of laser atmospheric propagation performance evaluation dataset

View table

View in Article

Table 2. Data range of laser atmospheric propagation performance evaluation dataset

Parameter	Data range
Propagation distance L/m	1000-20000
Atmospheric coherence length r₀/cm	1.4-290
Thermal distortion parameter N_D	0.002-500
Atmospheric transmittance T	0.08-1
Laser wavelength $ \lambda $/μm	1064
Propagation elevation angle θ/(°)	0-90
Platform height H/m	0-20
Laser power P₀/kW	30-500
Aperture D/m	0.72-1
63.2% extension radius ER63. 2%/cm	3-150
63.2% expansion multiple EM63.2%	4-66
83.9% extension radius ER83.9%/cm	4-160
83.9% expansion multiple EM83.9%	5-57

Table 3. Optimal hyperparameters of the beam quality evaluation factor prediction model
View table
View in Article
Table 3. Optimal hyperparameters of the beam quality evaluation factor prediction model
Beam quality evaluation factor d α
63.2% extension radius 3 2.0
63.2% expansion multiple 4 1.0
83.9% extension radius 4 1.5
83.9% expansion multiple 5 1.5

Table 4. Comparison of output results among MLR, SVM, and Lasso
View table
View in Article
Table 4. Comparison of output results among MLR, SVM, and Lasso
Model r R² E_MS E_MA E_MR
MLR 0.944 0.892 2e-3 0.021 51.12%
SVM 0.995 0.990 6e-5 0.006 5.36%
Lasso 0.999 0.999 9e-6 0.002 1.25%

Table 5. Comparison of computational efficiency among MLR, SVM, and Lasso
View table
View in Article
Table 5. Comparison of computational efficiency among MLR, SVM, and Lasso
Model Train time Inference time per sample
MLR 4.42 μs 0.13 μs
SVM 0.38 s 12.61 μs
Lasso 4.73 s 4.14 μs

Table 6. Average response time for a single piece of data for the Lasso regression model
View table
View in Article
Table 6. Average response time for a single piece of data for the Lasso regression model
Beam quality evaluation factor Time/μs
63.2% extension radius 4.14
63.2% expansion multiple 4.35
83.9% extension radius 4.19
83.9% expansion multiple 3.95

Table 7. Comparison of prediction errors for beam quality evaluation metrics
View table
View in Article
Table 7. Comparison of prediction errors for beam quality evaluation metrics
Beam quality evaluation factor E_MS E_MA E_MR
63.2% extension radius 9e-6 2e-3 1.25%
63.2% expansion multiple 0.23 0.35 3.52%
83.9% extension radius 2e-5 3e-3 1.47%
83.9% expansion multiple 0.28 0.40 3.97%

Tools

Get Citation

Copy Citation Text

Zhifu HUANG, Ying ZHANG, Nan LI, Xiaoxing FENG, Zhiqiang WANG, Chunhong QIAO, Chengyu FAN, Yingjian WANG. Research on a rapid evaluation method for laser atmospheric propagation based on machine learning[J]. Infrared and Laser Engineering, 2025, 54(7): 20250054

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Atmospheric optics and oceanic optics

Received: Jan. 15, 2025

Accepted: --

Published Online: Aug. 29, 2025

The Author Email: Nan LI (nli@aiofm.ac.cn)

DOI:10.3788/IRLA20250054

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Data structure at different levels of the Lasso regression model

Table 1. Data structure at different levels of the Lasso regression model

Table 2. Data range of laser atmospheric propagation performance evaluation dataset

Table 2. Data range of laser atmospheric propagation performance evaluation dataset

Table 3. Optimal hyperparameters of the beam quality evaluation factor prediction model

Table 3. Optimal hyperparameters of the beam quality evaluation factor prediction model

Table 4. Comparison of output results among MLR, SVM, and Lasso

Table 4. Comparison of output results among MLR, SVM, and Lasso

Table 5. Comparison of computational efficiency among MLR, SVM, and Lasso

Table 5. Comparison of computational efficiency among MLR, SVM, and Lasso

Table 6. Average response time for a single piece of data for the Lasso regression model

Table 6. Average response time for a single piece of data for the Lasso regression model

Table 7. Comparison of prediction errors for beam quality evaluation metrics

Table 7. Comparison of prediction errors for beam quality evaluation metrics