Carbon dioxide (CO2) as the important raw materials of plant growth in greenhouse, it is also one of the main factors that affects the plant photosynthesis. Adding CO2 gas fertilizer has been one of the important techniques for increasing production of tomatoes in greenhouse. In order to determine the proper amount of CO2 based on the plant demands, the impact of different CO2 concentrations on net photosynthesis rate (Pn) of tomato plants was studied. Tomatoes after planting were treated under four different CO2 concentration levels, including elevated CO2 concentrations of (700±50) μmol/mol (C1), (1 000±50) μmol/mol (C2), (1 300±50) μmol/mol (C3)， and ambient CO2 concentration in greenhouse (about 450 μmol/mol, CK). The above-mentioned CO2 enrichment was taken in the sunny daytime (09:00—12:00). During the experiment, firstly, the sensor nodes based on WSN were used to monitor greenhouse environmental parameters, including air temperature, air humidity, light intensity and CO2 concentration. Secondly, the diurnal dynamics of photosynthesis rate of tomato plants were achieved by LI-6400XT portable photosynthesis analyzer at the flowering stage. The parameters were measured by hourly from 08:00 to 18:00. In the environmental factors nested test of photosynthesis, the CO2 concentration gradients were set to 400, 600, 800, 1 000, 1 300 and 1 500 μmol/mol, respectively, the PAR gradients were set to 300, 600, 900 and 1 200 μmol/(m2·s), respectively, and the temperature gradients were set to 28℃ and 35℃, respectively. The air humidity came from the ambient environment (23.16%~46.07%). Then, in order to better understand the characteristics of tomato growth and achieve the purpose of the regulation of CO2 concentration in greenhouse, BP neural network was used to create photosynthesis prediction model according to the environmental factors nested test of photosynthesis. The diurnal dynamics of photosynthesis rate from different groups were simulated based on established model in the natural environment (except CO2 concentration), from which the CO2 concentration saturation point was obtained. The results indicated that CO2 enrichment raised Pn of tomato significantly, and the value was 37.13% and 40.42% higher in C2 and C3 than that in CK, respectively. Furthermore, the photosynthesis prediction model created by training group was accurate with average relative error of 3.91%, mean absolute error of 0.51 μmol/(m2·s), root mean square error of 0.79 μmol/(m2·s) and correlation coefficient of 0.98. The corresponding values of testing group were 10.08%, 1.36 μmol/(m2·s), 1.80 μmol/(m2·s) and 0.93, respectively. The prediction effect of diurnal dynamics of photosynthesis revealed that the correlation coefficient between the measured and calculated values was 0.96 when CO2 concentration was set to 700 μmol/mol, 0.94 when CO2 concentration was set to 1 000 μmol/mol, 0.78 when CO2 concentration was set to 1 300 μmol/mol, 0.96 in the 450 μmol/mol treatment. Therefore, the prediction model had high accuracy and certain universality, which could provide a theoretical basis for optimal regulation of photosynthetic rate dynamically and precise control of CO2 gas fertilizer in greenhouse.