摘要
快速S变换(FST)是解决实时监测电能质量问题途径之一,离散正交S变换(DOST)也是一种冗余度小的变换且被用于电能监测中。为了比较两种方法在实时电能质量扰动识别中的异同,首先提出一种快速实现DOST的方法,论证其计算复杂度与FST等同;比较常用的七种电能质量干扰信号在不同时窗下的FST系数和DOST系数,发现DOST模系数和采用sinc窗的FST模系数在每一个频段内比值呈线性关系,并且和采用高斯窗的FST模系数频谱结构类似,进而把这两种方法与支持向量机结合对电能质量干扰信号进行分类比较。仿真识别结果显示,快速DOST方法和FST在电能质量识别的精度和速度是一样的,在时间和准确性方面都比基于S变换的方法有一定的优越性。
Fast S-transform (FST) is an approach to the problem of real-time monitoring of power quality. Discrete Orthogonal S-transform (DOST) is also a transform with small redundancy that can be used for power quality detection. In order to comparing the two methods in the real-time power quality disturbances, a method for the rapid implementation of DOST is firstly proposed. The computational complexity of fast DOST is proved to be equivalent to FST. Comparison based on seven kinds of power quality disturbance signal shows that FST coefficients using sinc window are in proportion with the DOST coefficients in every frequency band, and spectrum structure of DOST is similar to FST using gaussian windows. Fast DOST and FST are combined respectively with the least squares support vector machine to classify power quality disturbance signal. Results show that DOST and FST method have the same accuracy and speed. Both of them are preferable to the method based on S-transform in terms of time and accuracy.
出处
《智能电网》
2014年第6期22-27,共6页
Smart Grid
基金
国家自然科学基金项目(61271091)
陕西省教育厅科学研究计划项目(2013JK0997)~~