Calibration of thermal sensors using BP neural network and SVM

In this paper, the influences of measuring distance and ambient temperature on the measurement accuracy of thermal sensors are explored through experiment. The data collected during the experiment are analyzed and used to train two machine learning models, i.e., back propagation (BP) neural network and support vector machine (SVM), with different numbers of hidden layer nodes and activation/kernel functions. Then, the models with better performance metrics are selected to compensate the measuring error of the thermal sensor. The experimental results show that both the BP neural network and the SVM can significantly improve the accuracy of the thermal sensor.