基于RBF-GA算法的特高压线路复合绝缘子均压环优化

Optimization Design for Grading Ring of Composite Insulators in UHV Transmission Lines Based on RBF-GA

特高压线路棒形悬式复合绝缘子表面电位分布极不均匀,绝缘子高压端金具附近绝缘子护套表面局部电场强度很大,加速了绝缘子的老化。工程上通常用安装均压环的方法来解决该问题。均压环结构参数决定了其均压效果,为得到较优的均压环参数,需要对其管径、环径、罩入深度的优化设计进行研究。由于复合绝缘子的结构特点使得单元数量较多,求解耗时长,常规遗传算法应用于该课题具有优化过程耗时较长的问题,因此使用改进的优化算法对均压环参数进行优化。改进的RBF-GA混合算法综合了RBF神经网络任意函数逼近能力和GA算法的全局优化能力,是一种性能较好的优化方法。文中通过有限元法对复合绝缘子电场分布进行了数值计算,用RBF神经网络对复合绝缘子护套上最大电场强度值进行了拟合,最后用GA算法基于拟合值对均压环参数进行了优化。基于RBF-GA算法对均压环结构进行参数优化,求解速度较使用传统的遗传算法有大幅度提高,并得到了良好的优化结果,证明了该方法的有效性。

UHV composite insulator surface potential is very unevenly distributed, and there is an intense local electric field strength on the surface of the insulator sheath near high voltage insulator end fitting,accelerating the aging of insulators. The method of installing the grading ring to resolve the issue is commonly adopted in engineering. The effect of electric field distributing depends on the structural parameters of grading ring. In order to conclude optimized structural parameters, it is of great necessity to study the optimization of the tube radius, the ring radius and the projection from high voltage end fitting.In this paper, an improved genetic algorithm（GA） has been adopted because the traditional genetic algorithm is time consuming in grading ring optimization. Radial base function-GA（RBF-GA） is an effective method which combines the approximate ability of any continuous function with arbitrary precision of RBF network and the ability of global optimizing of GA. This paper calculates the distribution of electric field by using finite element method and uses RBF network to approximate the maximum electric field strength on the surface of the insulator sheath near high voltage insulator end fitting. Finally, this paper optimizes the structural parameters by using GA based on the fitting value. The optimization speed of grading ring structural parameters based on RBF-GA method significantly is accelerated with good optimization results,which proves the effectiveness of the method.