Greenhouse culture, as a high-yield and efficient production technology, has become an important means to increase crop yield and quality. Greenhouse environmental monitoring is a key practice affecting the yield and quality of greenhouse crops. Meeting the growth needs of crops while reducing the cost of environmental monitoring has become an urgent and difficult scientific problem. Sensor technology with the addition of computational fluid dynamics (CFD) technique is a new method for addressing this issue. However, this CFD approach generally cannot monitor the greenhouse environment in real time. This study aims to establish a virtual sensor system based on the external environment of a climate model to form a closed-loop network using a finite volume CFD code (ANSYS FLUENT 19.0). An experiment is conducted on a flower greenhouse, and a 3D transient CFD is implemented including a meshing and turbulence model in the MATLAB interface program. Furthermore, a temperature model based on the external greenhouse environment is established using a regression method from the model, and a virtual sensor model of greenhouse temperature is established using a feedback method. The results show that the simulation satisfied the temperature distribution law, revealing that virtual sensors and physical sensors can work together to monitor the greenhouse environment.