The existence of a large number of harmonics in the power network seriously affects the safe and stable operation of the power system. Fast Fourier transform algorithm is widely used in the detection of harmonic in the power network. Because of the spectrum leakage and fence effect, the error of harmonic parameter detection is large. The precision of FFT algorithm can be improved by windowing function and interpolation algorithm. The sidelobe performance of window function is improved by multiplication and convolution.The Blackman self-multiplication convolution window is obtained by multiplying and convolution of Blackman window. The window function has better main lobe and sidelobe performance.The interpolation correction formula of frequency, amplitude and phase is derived by using the interpolation algorithm of triple-spectral-line.Using this window function and other windows simulate the complex signal with weak amplitude signal, it is verified that the triple-spectral-line interpolation algorithm of Blackman self-multiplication convolution window still has high accuracy in detecting weak amplitude signal. The simulation of the signal with white noise verifies that the algorithm has small relative error and strong anti-jamming ability to the harmonic signal parameter detection.