Abstract: Due to the high cost of greenhouse lighting control, the optimal control model of greenhouse light with benefit-priority was proposed to reduce the cost of regulation and control under the condition of meeting the needs of crop growth. A nested experiment was designed to collect the photosynthetic rate of Cucumber under different temperature, CO₂ and PPFD in greenhouse. Construction of photosynthetic rate prediction model based on LS-SVM with environment factor as input and photosynthetic rate as output. Then, the optimization algorithm which combines the two different schemes of light optimization is used to build the light optimization model based on BP neural network. The data shows that the theoretical light supply and demand decreased by 22.86% and the photosynthetic rate decreased by 3.71% compared with the control based on the light saturation point. In the validation test, the light supply and demand decreased by 27.47% and the photosynthetic rate decreased by 3.34%, which is 40.69% higher than that of the control group under natural conditions. It is of guiding significance for the precise control of light supplement in greenhouse.