Greenhouse agriculture has become an essential part of modern agriculture. The proper temperature is essential for greenhouse crops. Therefore, it is necessary to determine an appropriate greenhouse temperature automatic control strategy for greenhouse environment control. A model-free adaptive greenhouse temperature control strategy is proposed by combining a linear extended state observer and a fuzzy controller. Specifically, the linear extended state observer is introduced to estimate model approximation error generated by pseudo gradient estimation. The fuzzy controller is used to correct the system parameter online. In the absence of disturbances, compared with the typical model-free adaptive scheme, the MAE and RMSE of the proposed scheme are reduced by 57.55 % and 5.40 %, respectively. In the presence of custom disturbances, the proposed control scheme still has a faster response speed, stronger robustness, and higher control accuracy. And the proposed method can reduce the frequent action of the actuator, thus effectively reducing the energy cost.