Based on the minimum control cost model of traditional under-frequency load shedding（UFLS） parameter optimization， an optimization model of UFLS parameters is developed. This model considers the frequency recovery performance of the system and the minimum load shedding amount， together with the factors of risk probability. Based on the characteristics of the developed model， differential evolution（DE） algorithm is used to solve the problem. According to the steady constraint and overshoot constraint in optimization problem， piecewise punishment function method is proposed to improve the execution efficiency of DE algorithm. The practical example simulation of the power system shows that the comprehensive performance of the proposed model is better than minimum control cost model， feasibility and effectiveness of the proposed model and algorithm are validated preliminary.