The invention discloses a high-precision harmonic parameter estimation method based on sliding window DFT (Discrete Fourier Transform), which is suitable for estimating the harmonic components of a power system. After SWDFT is carried out on a power grid signal containing harmonic components in the power system, the transformation result has the same linear relationship as the original time domainsignal, namely, the transformed sequence still retains the mathematical structure of a multi-frequency signal, so that parameters can be estimated by using the existing harmonic analysis model. In addition, SWDFT enhances the signal-to-noise ratio of each component, and suppresses the interference of noise on the estimation performance. In the method, SWDFT is carried out on a harmonic signal inthe time domain first, and then, a Prony algorithm is adopted for the transformation result to obtain the estimated values of the frequency, amplitude and phase of each harmonic component of the signal. In order to enhance the estimation performance and robustness of the algorithm, a complex least squares (PLS) criterion is introduced to expand the original Prony algorithm. Therefore, the precision of parameter estimation is improved.