Machine Learning Assisted High Precision Vector Bending Sensor Based on Remodulate LPFG

Vector curvature fiber sensors have significant applications in many fields. Traditional fiber sensors rely on single peak demodulation, which often leads to inaccurate demodulation results. In this letter, a high-precision vector bending fiber sensor named Remodulate long-period fiber grating(Remodulate LPFG) is designed. We use the Residual multilayer perceptron model, which fully utilizes the information of multiple modes in the full spectrum to predict vector curvature. The results of the experiment show that the prediction accuracy is 99.93% with a mean absolute error (MAE) of 1.8° for bending direction measurement and 98.92% with an MAE of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.04~m^{-1}$ </tex-math></inline-formula> for the curvature measurement. Our experiments demonstrate that our model has high precision prediction. The high precision prediction and compacting structure consume Remodulate LPFG can unleash enormous value in engineering applications.