With the continuous development of demand response technology, the flexible loads of user side have shown more schedulable potential. Nevertheless, due to external factors such as user habits, model accuracy, and usage environment, most distributed loads reflect uncertainty, which generates difficulties to accurate load scheduling. Compared with the traditional box type uncertain set, the polyhedral uncertainty set introduces the load correlation coefficient when describing the sample uncertainty space, compressing the sample space uncertainty measure and improve the accuracy and economy of the optimization strategy. By analyzing the performance of the box type uncertainty set and the polyhedral uncertainty set under different correlation conditions, it is known that the load group scheduling optimization strategy model based on the polyhedral uncertainty set can reduce the scheduling cost, identify the high?quality response resources, and reflect the complementarity of negative correlation load.