A Multivariable Sensing System for Condition Monitoring of Oil-Immersed Transformers

Transformers are critical to electrical grids, making continuous monitoring of their condition essential. Transformer health is conventionally monitored through oil quality. Advances in monitoring can be achieved by focusing on quantifying species released during oil degradation. However, no existing technology quantifies these degradation markers simultaneously. Our goal was to develop sensing technology for these markers that meets key criteria: coexistent quantification and compatibility with transformer systems. We present a high-performance microwave sensor designed for efficient oil quality signal acquisition, coupled with a novel computational model for simultaneous quantification of degradation markers. Modified coaxial cable-based resonator working on multivariable response principles acts as a sensing device. It guarantees concurrent information about the oil and paper insulation conditions by quantifying moisture and 2-furfuraldehyde (2-FAL) simultaneously. Sensor performance is realized in terms of sensitivity, selectivity, repeatability, and limit of detection studies. Multivariable sensors are compact alternative to sensor arrays with reduced drift, minimized size, and lower costs. The use of microwave sensors for multivariable sensing is novelly reported, and their application in a transformer monitoring scenario is paramount owing to high stability. The proposed computational model can significantly facilitate the discriminative power of physical multivariable sensors by mitigating their selectivity issue.