Compound electric field prediction method for the UHVDC transmission lines based on fuzzy clustering and least squares support vector machine

作者：Deng, Jun^*; Xiao, Yao; Hao, Yanpeng; Li, Licheng; Zhao, Yuming; Zhang, Jiangong

来源：High Voltage Engineering, 2013, 39(12): 2987-2993.

DOI：10.3969/j.issn.1003-6520.2013.12.021

• Electromagnetic environments
• LSSVM
• Particle swarm
• Total electric fields
• UHVDC
• Wavelet transformations

We proposed a least squares support vector machine method (LSSVM) in order to predict the total electric field of UHVDC transmission lines in given atmospheric environment. The LSSVM is based on in-filed data extracted by using the fuzzy clustering method in the given environment, as well as the penalty coefficient and the scale factor of wavelet kernel function. Both the penalty coefficient and the scale factor are optimized by using the particle swarm algorithm. Practical calculation reveals that the effectiveness of the LSSVM by optimizing the penalty coefficient and scale factor with genetic methods is superior to that by particle swarm algorithm. The prediction obtained by the proposed fuzzy clustering method and LSSVM has a maximum average relative error of 6.43% when test samples values from half-voltage of positive pole and full-voltage of negative pole, full-voltage of bipolar at different altitudes are used. Therefore, for the total electric field prediction of the given parameters, the accuracy and efficiency of the proposed method is assured, thus the method can be used to analyze compound electric fields in UHVDC project environmental evaluation and transmission line designs, etc.

• 出版日期2013-12
• 单位南方电网科学研究院有限责任公司; 华南理工大学