In protected agriculture, extreme temperatures can cause irreversible damage to crops, making temperature regulation a critical component of greenhouse environmental management. The dynamic optimization of the temperature ranges serves as the core strategy to enhance production stability. Consequently, identifying the optimal temperature ranges is pivotal for maximizing greenhouse production efficiency. This study proposes a novel method for determining the optimal regulation ranges throughout the multiple growth stages of greenhouse-grown peppers, incorporating curvature theory. A nested experiment was designed to obtain the photosynthetic rate (Pn) of peppers during the multiple growth stages under variable temperature, CO₂ concentration, and photosynthetic photon flux density. A photosynthetic rate prediction model was then constructed using a backpropagation neural network optimized by a genetic algorithm, with an R² of 0.9812 and an MSE of 1.35 μmol/(m²·s). The prediction model was subsequently discretized and applied to calculate the Gaussian response surface of Pn. Finally, the U-chord algorithm and the random restart hill-climbing method were employed to precisely define the boundaries of the temperature regulation ranges. Practice demonstrated that the average dry weight of pepper fruits in the experimental group was 96.83% higher than that of the no-operation regulation group and 243.65% higher than the fixed threshold group. This method not only enhances pepper growth but also exhibits superior regulatory tolerance. Its innovative temperature regulation strategy provides crucial technical support and establishes a reliable decision-making basis for the precise environmental management of greenhouse crops in protected agriculture.

Automated and intelligent light supplement systems have been widely applied in greenhouses in recent years. Among them, the light environment regulation model can be the core content of the system. However, the existing models cannot consider the comprehensive influence of light quality and light intensity, as well as the double optimization of net photosynthetic rate and light use efficiency. In this study, a collaborative control method was proposed for the light quality and light intensity using multi-objective optimization, particularly for the efficient supplemental illumination of cucumbers in greenhouses. Firstly, a multi-factor coupled photosynthetic experiment was designed to obtain the net photosynthetic rate of cucumber leaves. The model of net photosynthetic rate was then established using support vector regression with temperature, carbon dioxide concentration, photosynthetic photon flux density and light quality ratio as the input, while the net photosynthetic rate as the output. Furthermore, the light use efficiency was calculated at the leaf scale, according to the definition. Secondly, a multi-objective optimization model was constructed with the light use efficiency and net photosynthetic rate as optimization targets, while the light quality and light intensity as control variables. The non-inferior solution set was solved using the multi-objective particle swarm optimization. Technique for order preference by similarity to ideal solution was used to select the control single point of light quality and light intensity, in order to narrow the regulation interval for the less subjectivity of manual selection. Finally, the red and blue light demand were calculated according to the multiple relationship of light quality and light intensity. And then the red and blue light models were fitted by support vector regression with the temperature and carbon dioxide concentration as the input. The control experiments were carried out to compare with the fixed light quality supplement and the photosynthetic maximum supplement, in order to verify the superiority. The theoretical verification experiment showed that the net photosynthetic rate decreased by 21.39%, whereas, the light demand decreased by 59.40%, compared with the photosynthetic maximum supplement. The net photosynthetic rate increased by 3.66% and 9.69%, respectively, compared with the fixed light quality of 0.5 and 0.8. The practical verification experiment was also carried out to further verify the energy efficiency. The results showed that the physiological indicators were better than the fixed light quality supplement under similar power consumption, indicating significant differences in the stem diameter, dry weight and strong seedling index. There was no significant difference in physiological indicators, but the power consumption decreased by 27.43%, compared with the photosynthetic maximum supplement. The consumption of light and electrical energy resources was effectively saved to keep the physiological indicators almost unchanged. The model construction can be expected to serve as the new perspective for the greenhouse light supplement. This study can provide a new light supplement strategy for the facility's agricultural regulation and the efficient utilization of agricultural production resources.

Facility agriculture is one of the most important indicators to achieve efficient and high-quality crop production in modern agriculture. Optimal environmental parameters can also be adjusted to improve crop growth, yield, and quality in the greenhouse. Therefore, environmental control and regulation technologies have been widely used to achieve efficient vegetable production. This review aims to summarize the recent research status in the field of environmental regulation in greenhouses. New approaches were also proposed for future research priorities. Greenhouse environment system was gradually shifted from the manual and timed to the threshold, feedback, and intelligent control systems, with the rapid development of artificial intelligence (AI) technology. Firstly, the main properties of different control systems were evaluated from an application perspective. Among them, the threshold control was simple and widely used, but it failed to adjust the control strategy in time following the dynamic changes of the external environment, in order to meet the needs of crops for light, water, and nutrients. The feedback control shared the stable environment through feedback regulation but was unsuitable for the complex multivariable conditions. Intelligent control was widely used to balance the interaction between different environmental factors in modern greenhouses. Afterward, the intelligent control methods were investigated for greenhouse environments, including fuzzy, decoupling, neural network, and environmental control, according to the crop phenotype parameters. Specifically, the mathematical model was independent of the controlled object in the fuzzy control, and easy to handle with nonlinear issues. However, the outline fuzzy was difficult to handle the sudden disturbances in the regulation system. In decoupling control, the appropriate control strategies were designed to transform the multiple parameters with coupling effects into a single variable. The regulation model was also constructed to integrate the multiple environmental factors and crop physiological needs. The intelligent control of the environment was realized in the development of greenhouse agriculture. The Neural networks were used to extract valuable information from a large amount of greenhouse environment data, thus providing powerful tools for the regulation models. The intelligent models mainly included single-factor, multi-factor, and multi-objective environment regulation. The data-driven method was one of the research hotspots in the intelligent regulation of greenhouse environments. However, the universality and economic benefits were the key limiting factors of regulation models. Efficient and accurate acquisition of phenotypic parameters greatly contributed to the fine management of greenhouse environments, indicating the intuitive, real-time monitoring, and dynamic regulation. However, it was still lacking in the interaction between phenotype and multiple environmental factors, which failed to apply directly in greenhouse production. In addition, the existing environmental control systems were evaluated for the light, temperature, air, ventilation, and irrigation greenhouse. Research directions were proposed to urgently improve and optimize the control system. Finally, future research and development trends were also recommended to construct the greenhouse environmental regulation, considering disturbance factors. Environmental regulation models were developed using crop growth and phenotype evaluation. A "cloud-edge-end" system of greenhouse environmental regulation was established to integrate multiple models. This finding can provide new ideas and references for the subsequent development of environmental control systems in greenhouses.