Capacitance State Evaluation of 750 kV Autotransformer Windings Based on BP Neural Network

Insulation structure changes, operational state variations, and internal defects in a 750 kV autotransformer can cause changes in the capacitance of the transformer windings, resulting in asymmetrical capacitance parameters of the three-phase windings and unbalanced foundation voltage on the low-voltage side winding bus. This paper offer a capacitance state evaluation fashion for 750 kV transformer windings based on BP neural network. By constructing a simulation model of a three-phase 750 kV autotransformer and considering the actual range of variation in winding capacitance parameters, a measurement dataset of unbalanced voltages on the low-voltage winding is obtained. The unbalanced voltage of the low-voltage winding and the winding capacitance are selected as input and output datasets, respectively. Based on BP and PSO-BP neural networks, transformer winding capacitance state evaluation models are established and trained. The capacitance state of the windings is evaluated and verified through simulation experiments exploitation unbalanced voltage data from a certain 750 kV transformer. The verification consequence show that the PSO-BP neural network model has better forecasting accuracy.