The problem of low accuracy in harmonic signal detection method of power system and mode aliasing in harmonic detection based on Empirical Mode Decomposition (EMD) is studied, a new method of harmonic detection based on Extreme-point Symmetric Mode Decomposition (ESMD) and Hilbert Transform (HT) is proposed according to the theory of ESMD algorithm. Firstly, the mode decomposition of the signal is carried out, and a series of Intrinsic Mode Functions (IMF) with different scale characteristics are obtained. Then, the instantaneous amplitude and instantaneous frequency of IMF are obtained by Hilbert Transformation (HT). This method can be used to decompose signal according to the characteristics of the signal itself. In theory, it has advantages of simple and high precision due to the expansion of IMF definition and internal interpolation method. The simulation results show that the method is adaptive in harmonic detection, better in accuracy and real-time. What’s more, it avoids the EMD aliasing in harmonic detection without adding noise. This work is supported by Natural Science Foundation of Shanghai (No. 15ZR1418000).