Abstract: In the greenhouse environment control, it is hard to gain the control object value correctly to take expert empirical set point as control object traditionally, because of the limitation of expert experience and the adaptability of crop growth, thus the greenhouse produce efficiency is affected. This paper presented a control method for the greenhouse environment, which can automatically set the environment control object according to the crop growth response by simulation. With this method the control objective value can be set automatically by model of greenhouse environment change and the crop growth using genetic algorithm, and OS-LSSVMR(online spare least square support vector machines regression) internal model control with good stability and robustness was adapted to carry on greenhouse environment control. The simulation results show that the method requires less energy, and more dry matter can be gained, It can gain more precision control compared with the traditional control method. Therefore, the method has the characteristics of energy saving and good stability, and can enhance the economic effectiveness of greenhouse production.