Accurate Strain Extraction via Kernel Extreme Learning Machine for Fiber Bragg Grating Sensor

Abstract

A novel machine learning based FBG sensing interrogation method is proposed in this paper. The method uses a Kernel Regularized Extreme Learning Machine optimized by Quantum Particle Swarm Optimization (QPSO-KRELM). The proposed method is divided into two steps. First, the prediction model is established through the KRELM. Second, the hyperparameters of the model are optimized by the usage of the QPSO algorithm. Due to the good stability and convergence of the QPSO-KRELM, the well-trained detection model can automatically and accurately extract the sensing information from the FBG spectrum. The experimental results demonstrate that the proposed method is reliable and efficient in interrogating the wavelength shift. Moreover, even if the FBG spectrum has few sample points, the method can still provide high demodulation accuracy.