论文部分内容阅读
为了计算给定大气环境参数下的特高压直流输电线路合成电场分布,基于模糊聚类方法提取了与给定环境参数相似的现场测试样本数据,利用粒子群方法优化惩罚系数和小波核函数的尺度因子,提出了特高压直流输电线路合成电场预测的最小二乘支持向量机(LSSVM)方法。并通过实际算例表明:采用粒子群方法优化惩罚系数和尺度因子的LSSVM方法计算效率优于采用遗传算法优化惩罚系数和尺度因子的LSSVM方法;基于模糊聚类和LSSVM方法的预测结果分别与正极半压和负极全压、不同海拔高度的双极全压合成电场测试样本的最大平均误差为6.43%。因此该方法应用于给定环境参数的特高压直流输电线路合成电场预测有着可靠的精度和效率,为特高压直流输电工程环境评估、线路设计等提供了有用的合成电场分析方法。
In order to calculate the electric field distribution of UHVDC transmission line under a given atmospheric environment parameters, field test sample data similar to the given environmental parameters were extracted based on fuzzy clustering method, and the particle swarm optimization method was used to optimize the penalty coefficient and wavelet kernel function scale Factor, a least square support vector machine (LSSVM) method is proposed for the electric field prediction of the UHVDC transmission line. The practical examples show that the LSSVM method using particle swarm optimization method to optimize the penalty coefficient and scale factor is better than the LSSVM method which uses genetic algorithm to optimize the penalty coefficient and scale factor. The prediction results based on fuzzy clustering and LSSVM are respectively compared with the positive The maximum average error of half-voltage and negative total voltage and bipolar total voltage synthetic electric field test samples at different altitudes is 6.43%. Therefore, this method has reliable accuracy and efficiency in predicting the electric field of EHV transmission lines with given environmental parameters, and provides a useful synthetic electric field analysis method for EHV HVDC project environmental assessment and circuit design.