The work aims to help designers grasp the potential connection between product color attributes and user image perception, and improve the efficiency of product color design. Multi-dimensional statistics and cluster analysis were used to obtain representative color samples and vocabulary samples under the PCCS color system, and color semantic quantification experiments were set up to obtain image ratings of color on various vocabulary scales. Based on the obtained evaluation value quantitative data combined with grey relational analysis, the color design evaluation model of the product was established in the Lab uniform color space through color transposition. Taking the two-color children's toy car as the object for product color design practice, the constructed color design evaluation model achieved the objectives of color screening, confirmation, evaluation and optimization of product color schemes. Combined with grey relational analysis, the product color design evaluation model constructed under the PCCS color system can effectively improve the color design efficiency of the product scheme, and provide an intuitive and accurate reference standard for the optimization of the product color scheme.