Hydroponic agriculture, as a modern agricultural technology, not only enables crops to better realize their growth potential but also frees them from the traditional agricultural constraints of land resources. Currently, the nutrient solutions for hydroponic crops are mostly prepared using generic formulas, lacking precise ratios based on the specific needs of different crop types and growth stages. It is of great significance to study the precise nutrient requirements of specific crops at different growth stages to enhance yield. To this end, a multi-strategy improved Grey Wolf Optimization algorithm is proposed in this paper to optimize the fertilizer model for hydroponic lettuce at different growth stages, providing more accurate nutrient requirements for hydroponic lettuce at different growth stages. Firstly, to solve the problems of the traditional GWO algorithm, such as insufficient exploration and the tendency to fall into local optima, non-linear functions, Symbiotic Organisms Search (SOS), and Opposition-Based Learning (OBL) were integrated into the GWO algorithm. Secondly, the objective was to minimize the residual function of the nitrogen, potassium, and calcium ternary fertilizer effect equation, the proposed algorithm was used to optimize the coefficients of the equation. Lastly, the optimal concentrations of NO3−, K+, and Ca2+ for each growth stage of lettuce were calculated based on the optimized fertilizer effect equations. The proposed algorithm was compared with five other excellent metaheuristic algorithms in optimizing fertilizer model coefficients. Experimental results indicate that the proposed algorithm outperforms the other algorithms used in the experiments. For the nutrient models of three growth stages, the proposed algorithm achieves improvements compared to the GWO algorithm, with R2 increasing by 0.73 %, 0.66 %, and 0.51 %, and RMSE decreasing by 8.53 %, 5.70 %, and 3.83 % respectively. This contributes to providing more accurate nutrient requirements for hydroponic lettuce.
Read full abstract