It is of great theoretical significance and practical value to research the effect of the very fast transient overvoltage (VFTO) on power transformers in gas insulated substations (GIS). This thesis proposes a method to calculate the distributed parameter of the power transformer winding by Back-propagation artificial neural network. Based on this method, the computer program is developed to calculate the voltage distribution of the power transformer. In the program, the unequal-length multiconductor transmission-line (MTL) model is adopted in considering the actual configuration of the transformer circular winding, and the compact finite difference (CFD) method is used to calculate the voltage distributions along transformer windings by means of vector fitting and recursive convolution to dispose the frequency-dependent parameters. For validation, the calculated results are compared with the measurements on the internal-shielded experiment transformer winding and the results are satisfactory. In addition, the voltage distributions of the 800kV and 1000kV high-voltage power transformer windings are calculated by the developed program under VFTO. |