Temperature compensation based on BP neural network with small sample data for chloride ions optical fiber probe

Fiber optic sensors have great applied in the field of sensing, however they are subject to temperature. In this study, we proposed an improved small sample data Back Propagating (BP) neural network for temperature compensation of a chloride ion probe based on an optical fiber Fabry-Perot interferometer (FPI). The temperature compensation results show that the Black Widow Optimization (BWO) algorithm was combined with BP neural network to further improve the performance of the model with a great detection accuracy that the relative error is 1.21%, associated with a Mean Square Error (MSE) of 2.6e−5.This is an excellent temperature compensation method with low computational cost and small samples.