Temperature prediction by helical microfiber sensors based on artificial neural network

Abstract

In this paper, we propose a back propagation neural network (BPNN) for temperature forecasting by helical microfiber sensors. The structural parameters, such as the microfiber diameter, the tapered angle, the input and output offset angle, the waist length and the helical angle, are considered as the input parameters of the network for sensing the temperature (T). 758 transmitted intensity (I)-T data pairs obtained from over 38 helical microfiber sensors are used for the network training. The prediction ability of the model is evaluated by root-mean-square error (RMSE). Compared with the fitting curve based on the measured I-T data, the neural network can directly predict the temperature according to the training model with RMSE of 0.6033.In addition, the major structural parameters are determined by comparing the prediction performances of the networks with different inputs.