Ming Yuan

In the process of agricultural production in solar greenhouses, the key to the healthy growth of greenhouse crops lies in accurately predicting environmental conditions. However, there are complex couplings and nonlinear relationships among greenhouse environmental parameters. This study independently developed a greenhouse environmental acquisition system to achieve a comprehensive method for the monitoring of the greenhouse environment. Additionally, it proposed a multi-parameter and multi-node environmental prediction model for solar greenhouses based on the Golden Jackal Optimization-Convolutional Neural Network-Bidirectional Gated Recurrent Unit-Self-Attention Mechanism (GCBS). The GCBS model successfully captures the complex nonlinear relationships in the greenhouse environment and accurately predicts changes in carbon dioxide concentration, air temperature and humidity, and soil temperature at different location nodes. To validate the performance of this model, we employed multiple evaluation metrics and conducted a comparative analysis with four baseline models. The results indicate that, while the GCBS model exhibits slightly higher computational time compared to the traditional Long Short-Term Memory (LSTM) network for time series prediction, it significantly outperforms the LSTM in terms of prediction accuracy for four key parameters, achieving improvements of 76.89%, 69.37%, 59.83%, and 56.72%, respectively, as measured by the Mean Absolute Error (MAE) metric.

Leaf photosynthetic pigments play a crucial role in evaluating nutritional elements and physiological states. In facility agriculture, it is vital to rapidly and accurately obtain the pigment content and distribution of leaves to ensure precise water and fertilizer management. In our research, we utilized chlorophyll a (Chla), chlorophyll b (Chlb), total chlorophylls (Chls) and total carotenoids (Cars) as indicators to study the variations in the leaf positions of Lycopersicon esculentum Mill. Under 10 nitrogen concentration applications, a total of 2610 leaves (435 samples) were collected using visible-near infrared hyperspectral imaging (VNIR–HSI). In this study, a “coarse–fine” screening strategy was proposed using competitive adaptive reweighted sampling (CARS) and the iteratively retained informative variable (IRIV) algorithm to extract the characteristic wavelengths. Finally, simultaneous and quantitative models were established using partial least squares regression (PLSR). The CARS–IRIV–PLSR was used to create models to achieve a better prediction effect. The coefficient determination (R²), root mean square error (RMSE) and ratio performance deviation (RPD) were predicted to be 0.8240, 1.43 and 2.38 for Chla; 0.8391, 0.53 and 2.49 for Chlb; 0.7899, 2.24 and 2.18 for Chls; and 0.7577, 0.27 and 2.03 for Cars, respectively. The combination of these models with the pseudo-color image allowed for a visual inversion of the content and distribution of the pigment. These findings have important implications for guiding pigment distribution, nutrient diagnosis and fertilization decisions in plant growth management.

Leaf photosynthetic pigments play a crucial role in evaluating nutritional elements and physiological states. In facility agriculture, it is vital to obtain rapidly and accurately the pigment content and distribution of leaves to ensure precise water and fertilizer management. In our research, we utilized chlorophyll a (Chla), chlorophyll b (Chlb), chlorophyll (Chll), and carotenoid (Caro) as indicators to study the variations in leaf position of Lycopersicon esculentum Mill. Under 10 nitrogen concentration applications, a total of 2610 leaves (435 samples) were collected using visible-near infrared hyperspectral imaging (VNIR-HSI). In this study, a "coarse-fine" screening strategy was proposed by using competitive adaptive reweighted sampling (CARS) and iteratively retained informative variable (IRIV) algorithm to extract characteristic wavelengths. Finally, simultaneous and quantitative models were established using partial least squares regression (PLSR). The CARS-IRIV-PLSR was used to create models to achieve a better prediction effect. The coefficient determination (R2), root mean square error (RMSE), and ratio performance deviation (RPD) were predicted to be 0.8240, 1.43, 2.38 for Chla, 0.8391, 0.53, 2.49 for Chlb, 0.7899, 2.24, 2.18 for Chll, and 0.7577, 0.27, 2.03 for Caro, respectively. The combination of these models with the pseudo-color image allowed for a visual inversion of the content and distribution of pigment. These findings have important implications for guiding pigment distribution, nutrient diagnosis, and fertilization decisions in plant growth management.