Decision method for greenhouse tomato light regulation based on the concavity of photosynthesis response

Light intensity is the key element of plant lighting, which determines the production efficiency of greenhouse. The photosynthesis response to light intensity was related to the environmental parameters and growth stages. First, based on the photosynthesis data obtained from the infrared gas analyzer, a photosynthesis prediction model that integrates multiple growth stages has been constructed using machine learning methods. Then, to determine the target values for light regulation under limited investment, the variation law of photosynthetic discrete value was analyzed by the central difference method, and the keypoints were extracted by the maximum concavity. Finally, to improve the applicability of the method, target values based on keypoints were calculated and modeled. The results showed that the mean absolute error and root mean squared error of the optimized photosynthesis prediction model were significantly reduced from 1.966 and 2.490 to 0.417 and 0.600, respectively. In addition, the first-order difference value was negative and monotonically decreasing, indicating that the increase rate of photosynthesis response function continued to decline, but this process was inhomogeneous. Based on the variation law, the average target values for the seedling, flowering and fruiting stages were 508.1, 457.4 and 462.3 mu mol & sdot;m-2 & sdot;s-1, respectively, which reduced the investment of light intensity by 67.79%, 69.56% and 70.07% compared to the light saturation points, with ROIs of 1.90, 2.02 and 2.13, respectively. The prediction models integrating growth stages is accurate, and the decision-making method significantly reduces investments while providing high returns.