Multiple incipient faults are practically known to exist in transformers. They tend to produce suddenly changing ratio limits in ratio‐based methods or oscillation of fault location in graphical methods. In consequence, the energy associated with them lies in‐between low and high severity single faults. Hence multiple fault detection needs to be addressed appropriately which may otherwise pose the serious constraints during transformer condition monitoring. In this study, novel and intelligent classification approach is proposed to upgrade the classical dissolved gas analysis (DGA) technique to cater the requirement of multiple fault diagnosis. This consists of Duval‐triangle‐based optimised fuzzy inference system and neural network models sensitive to both single and multiple incipient faults. Both models have been rigorously trained and tested using dataset credited to field and literatures to achieve high fault recognition and isolation rates, alternatively low false detection and no‐detection rates. Both parameters are combined into single index to determine the accuracy in terms of F 1 score which is evaluated to be >97%. The diagnostic ability of the scheme is highly promising and can improve reliability of transformer fault forecasting by DGA.

This article presents the novel application of a cross-capacitive sensor for monitoring insulation degradation of transformer oil. The condition of insulating oil is determined by measuring the variation of the moisture content and furfural (2-FAL) concentration in the range of 0–60 ppm. The moisture and 2-FAL are two important by-products formed due to the degradation of cellulose paper used for insulating transformer winding. In addition, a small concentration of moisture is always trapped in oil as well as in paper. An accurate and precise capacitive sensor is developed that uses the cross-capacitive structure proposed by Thomson and Lampard. The sensor is developed using bulk brass material and it consists of four identical brass electrodes that are separated by very small gaps. The sensor has been tested at different concentrations of moisture, and 2-FAL in four insulating oil samples that are commonly used in power transformers, and results are compared by a reference meter. The maximum accuracy of the sensor with respect to the commercial meter is 0.77% and the air cross-capacitance value differs calculated absolute capacitance value by 3%. The calibrated response shows a maximum error of 2%. The output of the sensor is highly precise (±0.0026% for moisture, ±0.0028% for 2-FAL), fast, drift-free, and the maximum nonlinearity of the response is close to 2%. The maximum standard deviations (SDs) for the measurement of the mineral oil moisture and 2-FAL concentrations are 0.025 and 0.023, respectively. The sensor response is independent of temperature variation.

A novel capacitive sensor is developed which can detect 2-Furfuraldehyde (2-FAL, C5H4O2, in ppm order) in transformer oil. No such sensor is reported earlier in the literature. The 2-FAL is released in transformer oil with the ageing of the transformer and is a promising indicator for remaining useful life of the solid insulation. The sensor is fabricated by coating a Molecularly Imprinted Polymer (MIP) made from Polydimethylsiloxane (PDMS) on a copper electrode. The MIP coating has the template of 2-FAL molecules. On testing, the 2-FAL molecules in transformer oil are adsorbed on the templates which result in a change of capacitance of the sensor. The sensor is tested in the range of 1-20 ppm for measuring 2-FAL concentration and shows an average sensitivity of 0.084 pF/ppm. The sensors are fabricated in different months (batches), and average capacitance values for different oil sample are presented. The sensor reliability is analysed using statistical analysis and repeatability study which shows repeatable results consistently.

On-site reliable transformer oil testing to assess the health of an oil-immersed transformer is a great challenge. Different sensors used for the purpose are having several limitations because of the oil and temperature. In this paper, a cross capacitance sensor cell is proposed for field testing of transformer oil which circumvents most of the limitations of thin-film chemical sensors used for the purpose. The cross-capacitance sensor cell based on the Thompson-Lampard theorem has been utilized for the first time to simulate a reliable and accurate sensor to determine degradation in insulating oil. The dielectric constant of the oil, which is determined for different oil samples experimentally, is an essential input for the finite element analysis of the sensor. Finite element analysis of the sensor cell is performed using ANSYS Maxwell software to determine its feasibility as a sensor to estimate the concentration of insulation degradation by-products dissolved in oil. The sensor is simulated to estimate the concentration of moisture and 2-FAL (2-Furfuraldehyde) in oil and to characterize aged oil. Finally, an experimental validation of the simulated sensor response for different fresh insulating oils is done. The experimental results closely match the simulation results. The sensor response is found to be approximately linear and very sensitive. Moreover, unlike thin-film chemical sensors, the same cross capacitance sensor can be used for the detection of different insulation degradation by-products dissolved in oil with proper calibration. Therefore, the proposed cross capacitance sensor cell can be used as a highly reliable on-site insulating oil degradation by-products detector.