With the rapid development of economy, the proportion of cooling load and heating load grows into an important share of the total load. These loads formed the peak load and had great demand response potential. The effective demand response strategy becomes greatly significant to achieve peak load shifting, load rate improving and electricity cost reducing, which is developed on studying users load characteristics in different seasons. Firstly, the daily load data in 2015 from Hainan power grid is divided into the different types according to the internal connection of the load curve by clustering algorithm, and the typical daily load curves are fitted out, which can represent the seasonal characteristics. Then the time-of-use (TOU) pricing model is established with the price demand elasticity coefficient to find the optimal price of different seasons. Finally, the users' power utilization under different TOU prices and different demand response strategies are compared. The result indicates that TOU price and incentive is significant for optimization, which is benefit for the development of intelligent power grid and demand response project.