Aiming at such problems as low accuracy and economy of traditional sintering distribution, an improved differential evolution algorithm for staged quadratic variation is proposed. Its characteristics: combining multiple mutation strategies, during the operation of the algorithm, the appropriate mutation strategy is selected in stages. When the fitness value is not updated for several consecutive iterations, adding a normally distributed perturbation factor can effectively improve the structure of the population, make the algorithm jump out of the stagnant state, and reduce the probability of falling into the local optimal solution. The adaptive values of each parameter in the algorithm are used to improve the performance of the differential evolution algorithm through eight standard test functions. The test proves that the improved algorithm has high precision and strong global optimization ability. Applying it to the sintering accessory problem can obtain a cost optimal solution on the basis of ensuring the basic composition of the compound.