Enhanced Temperature Sensing Performance of Optical Carrier-Based Microwave Interferometry With Artificial Neural Network

Temperature, as a crucial parameter in the fields of oil and gas, aerospace, and infrastructure condition monitoring, is commonly required for its prompt and accurate detection. In this article, a microwave photonics concept of optical carrier-based microwave interferometry (OCMI) is utilized for temperature sensing with the aid of the artificial neural network (ANN), which effectively improves performance. Different from the traditional dip frequency tracking demodulation method, the ANN-based demodulation method directly maps the microwave interferogram to the temperature, simplifying the demodulation process. With simple training, satisfactory temperature prediction accuracy can be obtained, which is not inferior to the traditional method. More importantly, the ANN-based demodulation method exhibits excellent performance in challenging situations such as limited sampling points and narrow microwave frequency bandwidths, significantly improving real-time response and easing equipment burden without compromising accuracy, which cannot be realized by the traditional method. Besides, it is revealed that the ANN-based demodulation method can enhance the performance of individual Fabry-Perot cavities, thereby facilitating the distributed sensing capability of OCMI-based interferometers.