基于s变换及遗传优化支持向量机的电压源换流器故障诊断技术研究

发布时间：2019-06-26 09:28

【摘要】：电压源换流器是VSC-HVDC系统最重要的部件,由于受过压和过流条件的制约,电压源换流器发生故障的可能性较高,维护较为困难。为了有效诊断电压源换流器故障,文中在VSC-HVDC系统电压源换流器故障特性分析的基础上,提出了一种基于s变换结合遗传优化支持向量机的电压源换流器故障分类方法。通过对直流母线电压幅频特性分析可知,故障信号各频率分量在低频段分布密集且主要集中在工频整数倍附近。据此,利用s变换提取故障信号低频段工频整数倍分量,构造模时频矩阵,采集各频率分量模值平方和及归一化模值平方和作为故障特征向量,利用遗传优化的二叉树多分类支持向量机实现故障类型的准确分类,并运用优化后的二叉树支持向量机对故障信号特征向量样本集进行训练和测试。结果表明,该方法不仅有效降低了特征向量维数,节约了计算时间,且能在样本较少的情况下达到较高的分类准确率。
[Abstract]:Voltage source converter is the most important part of VSC-HVDC system. Due to the restriction of overvoltage and overcurrent conditions, the possibility of voltage source converter failure is high and the maintenance is difficult. In order to effectively diagnose the fault of voltage source converter, based on the analysis of fault characteristics of voltage source converter in VSC-HVDC system, a fault classification method of voltage source converter based on s transform and genetic optimization support vector machine is proposed in this paper. Through the analysis of the amplitude-frequency characteristics of DC bus voltage, it can be seen that the frequency components of fault signal are dense in the low frequency band and mainly concentrated in the integer times of power frequency. Based on this, the integer multiple components of power frequency of fault signal in low frequency band are extracted by s transform, the mode time frequency matrix is constructed, the sum of squares of modules of each frequency component is collected as fault feature vector, the binary tree multi-classification support vector machine is used to realize the accurate classification of fault types, and the optimized binary tree support vector machine is used to train and test the eigenvector set of fault signal. The results show that the method not only effectively reduces the dimension of eigenvector, saves the calculation time, but also achieves high classification accuracy when the samples are small.
【作者单位】：国网福建省电力有限公司电力科学研究院;四川大学电气信息学院;
【基金】：国家电网公司科技项目支持(SGFJDK00DWJS1500058)~~
【分类号】：TM46;TM721.1

【相似文献】