Abstract: The problem of broadband resonance caused by high proportion of new energy grid connection is becoming more and more prominent. A broadband resonance early warning evaluation method based on inter-harmonics and clustering algorithm is proposed. Firstly, the detection principle of inter-harmonics is analyzed, and a fast Fourier transform ( FFT ) based on time window is proposed to detect zero-order inter-harmonics ( 5-45Hz ) and one-order inter-harmonics ( 55-95Hz ). Secondly, the big data clustering analysis algorithm is introduced to process the inter-harmonics, and the extracted data is subjected to K-means clustering analysis. Finally, the risk level of broadband resonance is comprehensively evaluated and analyzed. A simulation model with wind farm is built, and the effectiveness of the proposed early warning algorithm is verified by case analysis of the generated simulation data.