This study introduces a machine learning (ML)-assisted framework for optimizing Proportional-Integral-Derivative (PID) coefficients, enabling stable, long-length single crystal fiber (SCF) fabrication via the Laser Heated Pedestal Growth (LHPG) technique. Compared with traditional curve-fitting methods, the proposed ML approach significantly enhances computational efficiency and achieves up to 20dB improvement in prediction accuracy, leading to notable improvements in SCF uniformity and achievable fiber length. The fabricated sapphire SCF demonstrates a length exceeding 3 m and an average diameter of $150~\mu $ m with fluctuations below $\pm 3~\mu $ m, validating the effectiveness of the ML model. Additionally, fundamental guidelines for optimizing PID coefficients in LHPG systems are provided. These results offer valuable insights for high-quality SCF fabrication and are expected to facilitate the production of improved fibers suitable for harsh environment sensing applications.
Coherent diffraction imaging (CDI) is a fast-growing imaging technique. Among all CDI methods, coherent modulation imaging (CMI) has strong potential for dynamic imaging because of its ability to form an image from a single diffraction pattern. However, current CMI methods mostly reconstruct the exit wave distribution behind the object plane, which is seriously affected by the illumination artifact. Recently, some improved CMI methods have been developed to resolve the problem. However, many of these methods still need two diffraction patterns—one empty-sample diffraction pattern and another snapshot measurement. Recent advances in randomized probe imaging have shown that a single diffraction pattern suffices for quantitative reconstruction when the probe is pre-calibrated. Herein, we propose a modified CMI algorithm to reconstruct pure object function with single diffraction pattern, thereby simplifying the experimental process. Moreover, the proposed method can also work in the situation where the modulation effect is weak. Both numerical simulations and optical experiments have been conducted to verify the proposed method.
25 November 2025 in Journal of Lightwave Technology
In this study, we present and experimentally verify a method based on the adaptive matrix block weighting (AMBW) algorithm to achieve large-strain measurement in long-distance, high-spatial-resolution optical frequency domain reflectometry (OFDR) systems. The system generates two-dimensional (2D) cross-correlation images representing strain information by calculating cross-correlations between the reference signals (RS) and measurement signals (MS) for each segment of the sensing fiber, thereby determining the strain distribution along the fiber. In practical applications, however, strain-induced stretching causes spatial misalignment between the measurement and reference signals. Consequently, the cross-correlation images become contaminated with bad points and outliers, leading to a significant reduction in measurement accuracy. To mitigate the above effects and enhance measurement accuracy, we employ the AMBW method to process the 2D cross-correlation images. Unlike conventional image denoising techniques such as Wiener filtering (WF), Gaussian filtering (GF), and median filtering (MF), the AMBW algorithm leverages both the redundancy and inherent correlation within the 2D cross-correlation data by adaptively adjusting the size of the matrix blocks used in similarity weighting based on the spatial distribution of local outliers. This adaptive capability allows it to effectively eliminate bad points and outlier. Experimental results show that the proposed method accurately measures strain gradients from 500 μϵ to 5500 μϵ with a spatial resolution of 2 mm, operating at a wavelength scanning rate of 10 nm/s over a 56 m sensing fiber. The proposed approach offers an effective solution for long-distance, high-spatial-resolution, large-strain measurement in OFDR systems, underscoring its strong potential and practical value.