In many laser application scenarios, concentrated optical energy, high coherence, and narrow spectral linewidth are critical optical characteristics that ensure the excellent performance of lasers. These characteristics can be achieved when a laser operates in single longitudinal mode (SLM) rather than multiple longitudinal mode (MLM). Therefore, it is important to identify whether the laser operates in SLM or MLM accurately and efficiently, especially in scenarios with high real-time requirements such as high-precision time measurement. This study proposes a novel machine learning-based method for laser longitudinal mode identification, which has been effectively utilized in the development of an optical clock. Two machine learning classification models are designed, based on a support vector machine (SVM) and a convolutional neural network (CNN), respectively, with the datasets being the interference fringe data measured by a Fizeau wavemeter integrated in the optical clock. Using a dataset that includes 589 interference fringe samples from two different laser wavelengths, it is demonstrated that the machine learning models can achieve 96% to 100% classification accuracy in distinguishing between SLM and MLM. The methodology in this work offers valuable insights for future space missions that require high-precision measurements and lightweight payloads.
Yang et al. (Sun,) studied this question.