Prediction and Manipulation of Mode‐Locked Fiber Lasers Dynamics Powered by Spectral‐Spatial‐Temporal Mapping

ABSTRACT Mode‐locked fiber lasers (MLFLs) serve as critical ultrashort pulse sources and enable applications across multiple fields. Spectral analysis of mode‐locked pulses is essential to probe intracavity nonlinear effects and guide laser performance optimization. However, conventional spectral measurement techniques impose constraints on the repetition rate, mode field, and spectral resolution of the characterized lasers. Here, we introduce a high‐speed spectral characterization framework based on spectral‐spatial‐temporal mapping. By leveraging multimode intermodal interference, multicore fiber sampling, and convolutional neural network modeling, we demonstrate a MHz‐level high speed spectral analysis for an MLFL with a repetition rate of 8.63 MHz, achieving a high precision of 0.02 nm. Moreover, by incorporating automatic intracavity state optimization algorithms, the system enables intelligent laser spectrum manipulation without optical spectrum analyzer, confirming the robustness and practicality of the proposed methodology.