TY - JOUR

T1 - Intermittent Dynamics Identification and Prediction from Experimental Data of Discrete-Mode Semiconductor Lasers by Reservoir Computing

AU - Feng, S.D.

AU - Zhong, Z. Q.

AU - Her, Haomiao

AU - Liu, Rui

AU - Chen, Jianjun

AU - Huang, Xingyu

AU - Zhu, Yipeng

AU - Hong, Yanhua

PY - 2024/9/19

Y1 - 2024/9/19

N2 - Analysis of intermittent dynamics from experimental data is essential to promote theunderstanding of practical complex nonlinear systems and their underlying physicalmechanisms. In this paper, reservoir computing enabled dynamics prediction and identificationof two types of intermittent switching using experimental data from discrete-modesemiconductor lasers are rigorously studied and demonstrated. The results show that, for thedynamics prediction task, both regular and irregular intermittent switching can be predictedreliably by reservoir computing, achieving the average normalized mean-square error of lessthan 0.015. Additionally, the impact of the number of virtual nodes in the reservoir layer, aswell as the train-test split ratio on prediction performance, is explored. For the dynamicidentification task, a 2-class classification test is adopted, and the corresponding binaryaccuracy is calculated to evaluate the identification performance. The results demonstrate that,the accuracy of identifying both regular and irregular intermittent switching exceeds 0.996.Compared with the conventional amplitude threshold identification method, the reservoircomputing-driven dynamics identification method exhibits superior accuracy, especially in theintermittent transient transition regions

AB - Analysis of intermittent dynamics from experimental data is essential to promote theunderstanding of practical complex nonlinear systems and their underlying physicalmechanisms. In this paper, reservoir computing enabled dynamics prediction and identificationof two types of intermittent switching using experimental data from discrete-modesemiconductor lasers are rigorously studied and demonstrated. The results show that, for thedynamics prediction task, both regular and irregular intermittent switching can be predictedreliably by reservoir computing, achieving the average normalized mean-square error of lessthan 0.015. Additionally, the impact of the number of virtual nodes in the reservoir layer, aswell as the train-test split ratio on prediction performance, is explored. For the dynamicidentification task, a 2-class classification test is adopted, and the corresponding binaryaccuracy is calculated to evaluate the identification performance. The results demonstrate that,the accuracy of identifying both regular and irregular intermittent switching exceeds 0.996.Compared with the conventional amplitude threshold identification method, the reservoircomputing-driven dynamics identification method exhibits superior accuracy, especially in theintermittent transient transition regions

U2 - 10.1364/OE.538608

DO - 10.1364/OE.538608

M3 - Article

VL - 32

SP - 35952

EP - 35963

JO - Optics Express

JF - Optics Express

SN - 1094-4087

IS - 20

ER -