In response to the unreasonable preference coefficients of different weighting methods in the comprehensive weighting method and the problem of misjudgment in the maximum membership criterion, the rock brittleness coefficient σ_c/σ_t, rock mass stress characteristic coefficient σ_θ/σ_t, elastic energy index W_et, and rock integrity coefficient K_v are comprehensively selected as prediction indicators. By introducing a distance function to determine the preference coefficient of the improved Analytic Hierarchy Process and the Coefficient of Variation method, and combining the indicator set to obtain the comprehensive connectivity function. Finally, the maximum membership principle and confidence criterion are used for the comprehensive evaluation of rockburst intensity levels, and an improved rockburst intensity level prediction model based on comprehensive weighting and set pair analysis is established. Through the verification of 17 typical rockburst cases at home and abroad and actual engineering cases, it is shown that the proposed model has high accuracy and good practical application effects, which has certain engineering guidance significance.