基于改进递归DFT与SVR融合的实时谐波检测

Real-time harmonic detection based on combined algorithm of improved recursive DFT and SVR

根据电力系统中以整数次谐波检测为主要目标的情况,通过对递归离散傅里叶变换(DFT)和快速傅里叶变换(FFT)算法的分析比较,提出了改进递归DFT算法,即在递归DFT运算之前对首次采样做FFT运算,再将此运算结果作为递归DFT运算的初始值以进一步减少运算量。为同时提高检测精度,提出了基于改进递归DFT与改进序列最小最优化(SMO)的支持向量回归(SVR)融合的整数次谐波检测新方法。通过算例仿真,并与递归DFT及改进递归DFT相比较,证明了改进递归DFT与改进SMO的SVR融合的方法检测精度高,非常适合整数次谐波的实时检测和分析。

As the integer harmonics are mainly detected in power system,the improved recursive DFT （Discrete Fourier Transform） algorithm is proposed based on the analysis and comparison of FFT （Fast Fourier Transform） algorithm and recursive DFT algorithm,which carries out FFT of samples first,and carries out the recursive DFT with the result of FFT as the initial values to reduce the calculation time. In order to improve the detection accuracy,a new method of integer harmonic detection is put forward,which combines the improved recursive DFT with the SVR（Support Vector Regression） of improved SMO（Sequential Minimal Optimization）. Simulation results show that,compared with the recursive DFT and improved recursive DFT, the proposed method has higher detection accuracy,very suitable for the real-time detection and analysis of integer harmonics.