Abstract: In order to realize high-accuracy measurement of nonlinear loads, an improved algorithm of harmonic energy measurement based on discrete wavelet packet decomposition and reconstruction was proposed. Frequency band and the corresponding number of decomposition layer were given according to the sampling number, window width and sampling frequency. To improve the non-ideal cut-off characteristic of wavelet filter and eliminate the frequency aliasing, the discrete Fourier and the inverse transformation were used to remove the extra frequency components. The calculation expression of harmonic power and the principle of harmonic energy measurement were listed. The simulation analysis of harmonic energy measurement using db40 wavelet function indicated that the improved algorithm presented restrained frequency aliasing to a certain extent, and it could provide an effective method for accurately detecting harmonic, inter-harmonic and time-varying harmonic of electrical power system.

Key words: non-linear loads, harmonic energy measurement, discrete wavelet packet transforms, single-band reconstruction, frequency aliasing