A Study on Adaptive Robust Hybrid Algorithms for Fiber Optic Amplifiers and Their Application in Intelligent Parameter Identification

Abstract

To address the issues of low parameter identification accuracy in fiber-optic amplifiers under real-world noise environments, the limited performance of single algorithms, and the reliance of hybrid strategies on human experience, this paper proposes an adaptive robust hybrid algorithm framework. This framework integrates the particle swarm optimization (PSO) algorithm with the Levenberg-Marquardt (LM) algorithm, designs two hybrid modes-serial relay and embedded collaboration-introduces an adaptive rule scheduling mechanism based on population diversity and fitness rate of change, and employs the Huber loss function and data augmentation techniques to construct a noise-resistant system. Validation results on a simulated model of an erbium-doped fiber amplifier demonstrate that the proposed algorithm outperforms single-algorithm approaches in both convergence speed and accuracy. Furthermore, parameter estimation errors are significantly reduced in a noise environment with a signal-to-noise ratio of 25 dB. This approach provides a new paradigm of high precision and reliability for the intelligent modeling and robust identification of complex optoelectronic devices, and holds significant practical value for the digital twin and intelligent operation and maintenance of optical communication equipment.