There are multiple choices in the design of the target mix proportion and gradation composition of asphalt mixtures, and there is a problem of how to obtain the optimal gradation, which is related to the quality of mix proportion design and directly affects the pavement quality and service life of the pavement. Therefore, a scientific method is needed for gradation optimization. Based on the optimization theory, analyze the characteristics of mixed material performance indicators, and construct a three-level optimization index system consisting of non quantitative indicators, conventional quantitative indicators, and road performance indicators. Design three sets of typical gradation, conduct typical performance tests and analysis, and obtain the values of each evaluation index. Using Delphi expert survey method for membership degree, blind degree analysis, and overall understanding index weighting. Using grayscale model theory, the evaluation indicators are normalized, and the correlation degree of each indicator is analyzed. The grayscale model system is comprehensively evaluated and ranked step by step, and the optimal target mix ratio grading is ultimately selected based on the ranking results. In the study, the indicator system and process of the grayscale model analysis method were proposed, and the influence of the resolution coefficient values in the method on the optimization results was discussed. Research has shown that this method can systematically and comprehensively analyze the performance index system corresponding to the gradation composition, screen out the optimal characteristics that meet the road performance indicators, and provide an effective technical approach for the gradation optimization of asphalt mixture target mix design.