Regression Decision Trees Used With An Interferometric Sensor for Improved Temperature Measurement

Nowadays, machine learning techniques are used in classification and regression tasks. In this work, the decision tree method, which is a basic supervised learning algorithm, was used to predict the temperature of an interferometric optical sensor by means of a regression based on the characteristics of the reflecting optical spectrum. Generally, optical sensors have limited ranges in their measurements because they have ambiguities due to phenomena in the spectral range free of light interference. The spectral characteristics have some features such as wavelength and amplitude of the reflective spectrum of the fringes that present nonlinear relationships in response to the perturbation of the physical phenomenon being measured in this case temperature. It is concluded based on the results obtained throughout this work that by implementing the decision tree technique it is possible to overcome the ambiguities of the FSR, corresponding to 4 periods, and to extend the temperature measurement range (10 to 40 °C). Finally, it was possible to obtain an RMSE of 3.99×10-5 °C in the temperature prediction.