基于SPRGA遗传算法的高压直流分压器均压环结构优化设计

Optimization design of grading ring structure of HVDC voltage divider based on SPRGA

高压直流分压器外绝缘电场分布极不均匀,均压环的位置和尺寸参数是决定均压环表面场强大小的因素,如设计不当会存在最大沿面场强超过电晕起始场强的情况。针对均压环的最佳参数不易求解的问题,传统遗传算法存在局部收敛的现象,采用小种群实数遗传算法(small population real-coded genetic algorithm,SPRGA)和BP(back propagation)神经网络相结合的方法对高压直流分压器均压环进行了优化设计。计算结果表明,通过SPRGA和BP神经网络相结合优化后的均压环结构参数,可以减小均压环表面场强与沿面场强,减少电晕放电的概率,且相比传统遗传算法,其能进一步改善高压直流分压器电位与电场的分布,有利于提高直流分压器均压环设计的可靠性、实用性。

The external insulation electric field distribution of the high voltage direct current(HVDC)voltage divider is extremely uneven. The position and size parameters of the equalizing ring are the factors that determine the strength of the equalizing ring surface. If the design is improper, the maximum surface electric field strength may exceed the corona initial field strength. Aiming at the problems that the optimal parameters of the voltage equalizing loop are not easy to solve, the traditional genetic algorithm has the phenomenon of local convergence, the combination of the small population real-coded genetic algorithm(SPRGA) and the BP(back propagation) neural network is used to optimize the design of the voltage equalizing loop of the HVDC voltage divider. The calculation results show that the structural parameters of the equalizing ring optimized through the combination of SPRGA and BP neural network can reduce the surface field strength and along-surface field strength of the equalizing ring, and reduce the probability of corona discharge. Compared with the traditional genetic algorithm, this new method can further improve the distributions of the potential and electric field of the HVDC voltage divider, which is beneficial to improve the reliability and practicability of the direct current voltage divider equalizing ring design.