基于有限元和神经网络方法对220kV盆式绝缘子屏蔽罩结构优化设计  

Optimization of Shielding Electrode Design for 220 kV Basin-Type Insulator by Finite Element Method and Neural Network Method

作　　者：淡淑恒[1] 吴娜[1] 姜腾[2] 李昊东 高鹏[2] 朱锋[2] 王洪信[2] 

机构地区：[1]上海电力学院电气工程学院,上海200090 [2]国网淄博供电公司,山东淄博255000 [3]国网日照供电公司,山东日照276826

出　　处：《电瓷避雷器》2017年第3期153-158,165,共7页Insulators and Surge Arresters

基　　金：基金项目:上海绿色能源并网工程技术研究中心(编号:13DZ2251900)

摘　　要：针对GIS中盆式绝缘子,提出了一种优化其屏蔽罩结构参数的方法。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屏蔽罩结构优化目标;在此基础上,引入神经网络算法,利用BP神经网络拟合了屏蔽罩各结构参数与优化目标之间的关系,对屏蔽罩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,安装优化后的屏蔽罩,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。An approach to optimize the structure parameters of the shielding electrode is proposed for basin-type insulators in GIS. Firstly, this method establishes a model for basin-type insulators and calculates the electric field by the finite element method. Basing on the calculation results of the model, the BP neural network method is introduced to optimize the structural parameters of shielding electrode. In addition, the highly nonlinear mapping capability of neural network is used to fit the functional relationship between the shielding electrode structure parameters and optimization goal, which avoiding the problem of intensive calculation and tedious time consumption caused by the exhaustion method. The results show that, installing optimized shielding electrode can not only greatly lower the maximum value of surface electrical field, but also apparently improve the surface electrical field distribution of the basin--type insulators.

关 键 词：盆式绝缘子 电场计算 屏蔽罩 优化设计 有限元法 神经网络算法 

分 类 号：TM216[一般工业技术—材料科学与工程]