Abstract: At present, the problem of wind curtailment is serious in the "three north" areas of China. Installing heat storage tank（HST） in thermal power plant is one of the effective methods to absorb wind energy. In order to solve the problem of operation strategy of HST, an hourly operation model of combined heat and power（CHP） unit with HST system was established, and the operation strategy of HST was optimized by using improved genetic algorithm（GA） combined with characteristic day method. Results show that the optimal operation strategy is different with different objection function. When the maximum profit of CHP unit is taken as the objective function, the operation strategy must be optimized monthly according to the characteristics of heating load and power load in each month. The HST was used to store heat at high minimum power load and release heat at low minimum power load, which makes greater profits. When the maximum wind energy absorbed by system is taken as the objective function, the basic mode of storing heat at daytime and releasing heat at nighttime is followed. However, the capacity to absorb wind energy is related to the monthly heating load. When the heating load is high, the unit has a strong capacity to absorb wind energy. In addition, the optimal capacity of HST is the maximum of the optimal capacity in each month.