基于SVR的电能质量数据压缩算法

针对电力系统日益突出的海量数据流量的传输和存储问题,提出二维小波与支持向量回归结合算法用于电能质量数据压缩.利用小波变换把二维电能质量图像分解到不同尺度的子空间,对得到的不同方向的小波系数采用不同的数据组织方式.高频子空间系数采用可控制压缩比的ν支持向量回归(ν-SVR)学习系数间的相关性,用稀疏的支持向量表示原始数据,可以达到去冗余和数据压缩的效果.仿真实验利用不同的电能质量事件测试样本,对本文算法与传统支持向量机以及小波阈值法的压缩性能进行测试,结果表明,本文算法的压缩性能相比有了一定的进步.

A power quality event data compression algorithm base on support vector regression

Data storage and  communication currently pose a major problem for power quality and power systems monitoring, a method using 2 d wavelet and support vector machine for power quality event data compression was presented. First, 2 d representation power quality data was decomposed into wavelet  frequency subspaces. High frequency subspaces were compressed by νSVR, the coefficients’ correlation in wavelet domain was analyzed and represented by  sparse support vectors, therefore the original data.  could be compressed based on this feature. Experimental results showed that the compression performance of the algorithm achieve much improvement when compared to traditional support vector machine and wavelet algorithm．