基于有限元和神经网络方法对220kV盆式绝缘子均压环结构优化设计

Optimization Design of Grading Ring for 220 kV Basin-type Insulator Based on Finite Element Method and Neural Network Method

针对GIS中盆式绝缘子,提出了一种优化其均压环结构参数的方法。这种方法首先运用有限元方法对盆式绝缘子建立模型,并进行电场计算,然后以此确定了盆式绝缘子均压环结构优化目标;在此基础上,引入神经网络算法,利用BP神经网络拟合了均压环各结构参数与优化目标之间的关系,对均压环的结构参数进行了优化设计,得到最优结构参数,克服了传统穷举法计算量大、消耗时间长的缺点。实验结果表明,在盆式绝缘子上安装优化后的均压环,可以大大降低盆式绝缘子沿面最大场强,有效改善盆式绝缘子沿面电场不均匀程度。

A method for optimizing the structural parameters of the grading ring was proposed for basin-type insulator in GIS. Firstly, a model of basin-type insulator was estabhshed by adopting the finite dement method to calculate its electric field, hence the optimization objective of the grading ring structure was obtained. Then, the BP neural network algorithm was employed to optimize the structural parameters of the grading ring. In addition, the highly nonlinear mapping capability of the neural network is used to fit the optimization objective function with the structural parameters of the grading ring and goal, avoiding the huge calculation and large time consumption of the exhaustion method. Experimental results show that installation of the optimized grad/ng ring to the basin-type insulator can greatly lower the maximum value of surface electric field, and apparently improve the surface electric field distribution of the basin- type insulator.