Each separation system of multi-process steam coal preparation plant was analyzed and an optimization model of product structure was established. By translating distribution curve, product structure formulas under different separation density were obtained. Productivity and ash of primary product were predicted under the condition of different coal washing rate of raw coal, then economic benefits of raw coal per ton was calculated. Separating density and washing rate were obtained at the maximum economic benefits by using differential evolution algorithm, so as to get the optimum product structure. The application show that result calculated by the product structure optimization model for classified gravity separation in steam coal preparation plant based on differential evolution algorithm can meet with the actual one.