Temperature Optical Sensor Made of Recycled Waste Materials and Implementation of Machine Learning Method to Expand Its Measurement Range.

The efficient use of resources can contribute to the eradication of poverty and the reduction of environmental impact. Therefore, the use of recycled waste materials is desirable since it helps in the reduction of pollution. Some devices, such as interferometric optical sensors, can be implemented with this type of materials. In addition, optical sensors play an important role in the Internet of Things since they can be used to implement networks that simultaneously monitor several physical parameters. A problem that limits the measurement range of interferometric optical sensors is the 2π ambiguity. In this work, it is presented that the nominal measurement range of this type of sensor can be increased considerably by applying machine learning algorithms. In this sense, the kernel ridge regression (KRR) method is applied to the signals of a temperature sensor that is based on a tunable optical two-layer interferometer. Here, it is shown that this interferometric sensor can be fabricated with recycled waste materials (a polished stainless-steel plate, a single-mode fiber, a syringe needle, and epoxy clay), and the fabrication process is described. Furthermore, it is demonstrated that when KRR is used, the measurement range is increased at least two and a half times compared with that reached with traditional methods, and the temperature can be estimated with a root mean square error of 1.97 × 10-5 ℃. [ABSTRACT FROM AUTHOR]