Abstract: Abstract: In recent years, horticulture industry has been rapidly developed in China. The total area of cultivation is about 3.79 million hm2 by the end of 2012, but the climate control methods in actual greenhouse production are still relatively simple. Although many of the advanced intelligent control algorithms have been studied, however, there are two problems of these intelligent algorithms: the first one is the intelligent algorithms depend on the model of the greenhouse at the time of the operation. The control effect is better, only if the greenhouse model is accurate, but the greenhouse is a multivariable complex system with high coupling, so it is difficult to obtain accurate models. In addition, there are many different types of greenhouse in China, and each greenhouse structure may not be exactly the same, so it is inconsistent with the theoretical model. The second problem of intelligent algorithms is a large number of calculations are performed; the requirement is high for greenhouse controller’s processing capability and computing capability. The typical method for the climate control is to configure the static set point in actual greenhouse production, and the energy will be wasted because the static set point can not be automatically adjusted according to the external environment. In order to solve this problem, the greenhouse climate control method based on temperature integration was studied. When using this method, the first thing was to determine the expected average temperature, the maximum temperature and the minimum temperature in a certain period of time according to the type and growth stage of crop. Secondly, the actual average temperature of the current date would be calculated by the expected average temperature and the actual average temperature of previous day. Thirdly, the days were divided into N equal time intervals, and the length of each interval was int. The temperature set point of current time interval n would be calculated according to the actual average temperature and actual average temperature of previous n days, and the temperature set point contained 2 values: heating set point and cooling set point. Then the actual temperature of the time interval n and the temperature set point were compared. It would be heated if the actual temperature was lower than the heating set point, and the cooling would be operated if the actual temperature was higher than the cooling set point, otherwise no operation would be performed. Two comparative experiments were designed to verify this method using greenhouse simulation software, one experiment used the static set point to control the greenhouse temperature and the other experiment controlled it by temperature integration algorithm. In the greenhouse simulation program, a heater was for heating with the efficiency of 400 W/m2, a ventilation window for cooling, and the samples were collected once every 10 min which collected information such as time, internal temperature, humidity, heater working state and ventilation working state and a total of ten-day system simulation. The static set point temperature range was 16-25℃, the actual average temperature in greenhouse was 18.62℃ at the end of the simulation, and all of heating energy consumption was 167.39 GJ. The same temperature 18.62℃ was set as the expected temperature by the temperature integration algorithm for the control of greenhouse temperature. For the sake of obtaining the most solar radiation energy and reduce the heating energy, the maximum temperature was set to 30℃, the minimum temperature was 10℃ and the integral time was 2 days. The actual average temperature in the greenhouse was 20.21℃ and the total energy consumption for heating was 164.08 GJ by the temperature integration algorithm, and 98% of energy was used but the temperature of greenhouse was improved by 1.09 times through this method. The advantage of this method was proved by the more experiments in which the expected average temperatures were set to 17, 16 and 15℃, the actual average temperature in the greenhouse were corresponding to 19.67, 19.14 and 18.61℃ and the consumption of heater were 143.46, 126.07 and 107.85 GJ respectively. It can be seen by the results that the temperature integration algorithm has used only 64.43% of the energy but is able to achieve the same effect compared with the static set point method. Furthermore, the algorithm is simple and less calculation compared with the intelligent algorithms. Therefore, it can be used in common greenhouse controller and it has obvious energy-saving effect, but the greenhouse owners do not have to increase investment.