Construction of a tomato seedling growth model and economic benefit analysis under different photoperiod strategies in plant factories

Increasing demand for sustainable and efficient agricultural practices has driven the development of innovative production systems. Plant factories with artificial lighting (PFALs) serve as modern forms of facility-based agriculture and enable efficient crop production through precise environmental control. This study assessed the effects of different cultivation substrates (rock wool and coco coir) and photoperiod treatments (12 h/12 h, 16 h/8 h, and 20 h/4 h) on the growth and production efficiency of tomato seedlings in a plant factory. The results indicated that the bagged coco coir significantly enhanced water content and biomass accumulation in tomato seedlings by increasing the root zone temperature. The 20 h photoperiod markedly increased plant height, stem diameter, leaf area, and biomass compared to the 12 and 16 h treatments, without altering resource allocation. Growth curve analysis revealed that true leaf expansion was the critical phase in which growth differences became evident. Machine learning models based on multiple indicators [growth days, photoperiod, and cumulative light integral (CLI)] demonstrated that the Gradient Boosting Decision Tree (GBDT) model performed the best, with an R 2 value of up to 0.972. The feature importance analysis indicated that plant height and cotyledon development were primarily influenced by growth days, whereas CLI was key for predicting stem diameter and leaf area. Furthermore, the energy consumption and profitability analysis suggested that short to long photoperiod combinations (e.g., 12–16 h or 12–20 h) could reduce energy consumption while improving production efficiency, achieving the maximum annual output value. In conclusion, integrating machine learning to model tomato growth patterns, along with optimized substrate selection and lighting strategies, could effectively reduce energy consumption, enhance production efficiency, and promote sustainable agricultural development.