Nonlinear E-field Simulation of HVDC Converter Transformer Busing and Its Structure Optimization With RBF Neural Network and NSGA-II Algorithm

Static DC and PR are typical operation conditions of HVDC converter transformer bushing outlet device, moreover, the electrical parameters of RIP, transformer oil and OIP have the nonlinear relationship with E-field stress and temperature. Under the temperature gradient, the electrical parameters will modify the static DC and PR E-field distribution of the composite insulation. In order to simulate static DC and PR electric field of the composite insulation structure, this paper established the full model of the converter transformer outlet. Then, the structure parameters of converter transformer outlet have been optimized using combined RBF neural network and NSGA-II algorithm. Moreover, electric field verification calculation has been carried out under AC and PR conditions. The results show that under uniform temperature condition, E-field strength is larger than that under iso-thermal condition. The overshoot coefficient of the bushing end and barrier reach 65% and 45%, respectively. The study can provide reference for the design of outlet structure of the converter transformer.