Robust Sensor Selection for Reconstructing Thermal Properties in Electromagnetic Devices

Abstract

Electromagnetic devices have gained widespread use in various systems such as renewable energy systems, electrical motors, generators, and transformers. Despite the state-of-the-art modeling techniques, there are differences between the measured thermal behavior of electromagnetic devices and modeled ones. This research aims to bridge this gap by employing a combination of the finite element method and inverse modeling technique via non-collocated sensor configurations. Due to the restricted physical space and economic constraints, only a limited number of sensors can be strategically positioned within a structure. Consequently, the problem of robust and optimal sensor placement holds crucial significance on the accuracy and quality of the collected data influencing the performance, energy efficiency, and the measured thermal behavior of these devices. The objective of optimally locating sensors to acquire temperature data is to minimize the number of sensors and determine the optimal locations for capturing the most sensitive information. In this research, the challenge of robust and optimal sensor placement in the presence of uncertain thermal parameters is addressed using the Gramian-based method, facilitating the reconstruction of thermal properties by capturing the most sensitive temperature data. The experimental and simulation results demonstrate the effectiveness of the proposed approach in optimally selecting and placing thermal sensors and accurately determining the thermal parameters of the electromagnetic devices even in the presence of parameter uncertainties.