Abstract
We propose to use an extreme learning machine (ELM) to process sensing data of optical frequency domain reflectometer (OFDR) to overcome the disadvantage of long training time of the multilayer perceptron (MLP). The training set for machine learning is composed of training spectra, which are constructed by using Rayleigh scattering spectral models and reference spectra. After temperature experimental verification, the mean absolute error of temperature measurement by ELM is only 0.04, which is 78.6%, 69.2%, and 63.6% less than that of the traditional cross-correlation algorithm, MLP, and multilayer perceptron based on the sparrow search algorithm (SSA-MLP), respectively. The average data processing time of ELM is 0.17 seconds, and the training time is within 1 second, which is much smaller than that of SSA-MLP and MLP. It's illustrated that ELM is very competitive in achieving high accuracy and rapid temperature measurement, providing a new solution for temperature measurement in complex environments.
PDF Article
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