Utilising Genetic Algorithm and Driving Point Impedance Data to Synthesize High Frequency Circuit Model of Power Transformer Winding

Condition monitoring helps in achieving cost-effective and reliable power plant maintenance. Power transformer is an essential component of power system. To ensure its continuous operation, various condition monitoring techniques are being used. Frequency response analysis (FRA) is used for detecting severity of mechanical deformations in power transformer. In this paper, FRA is performed and driving point impedance (DPI) is recorded. The genetic algorithm (GA) is then utilised to synthesize a physically realizable, electrically and magnetically coupled lumped-parameter ladder circuit of a transformer winding from the measured DPI data. The fitness function and constraints are defined to narrow down the search and to obtain unique ladder circuit corresponding to the given FRA data. The validity and accuracy of the proposed method is verified by considering a circuit model of healthy transformer winding. The result shows that the synthesized circuit model is accurate. Further, terminal behaviour of the synthesized circuit is found similar to that of actual circuit model (representing healthy transformer winding). Thus, the method is reliable and can be extended for actual power transformer.