State estimation methods applied to transformer monitoring

State estimation methods are valuable for monitoring complex systems with not directly measurable states. One case in point is the monitoring of power transformers. The thermal condition of the transformer is very crucial in determining the life expectancy of the transformer, yet the hot spot temperature of the transformer cannot be directly measured. Other events such as inter-coil discharges cannot be directly measured unless they develop into a power fault. This paper describes the application of state estimation methods in a transformer diagnostic system. This system consists of hardware and software that continuously monitor the transformer electro-thermal states. The transformer states are obtained from redundant measurements of transformer terminal voltages, currents, tank temperature, oil temperature and ambient temperature. This data are fitted into the electro-thermal model of the transformer via state estimation methods that provide the full range of the transformer electro-thermal states. The paper describes the electrothermal model, the unique features of the data acquisition system and focuses on the application of the state estimation methods to derive important transformer quantities such as: (a) loss of life; and (b) transformer coil integrity. The system has been implemented on a trial basis on four Entergy System transformers. In the present implementation, the minimum input data requirements are: (1) phase currents and voltages at both ends of the transformer (a total of twelve); (2) top of oil and bottom of oil temperatures; and (3) load tap changer position. Actual data and the performance of the state estimation methods are presented.