Optimization of Nitrogen Fertilization and Planting Density for Enhancing the Leaf and Seed Performances of Woad Using Artificial Neural Network

Usually, effects of three or four levels of independent variables (here, nitrogen fertilizer and planting density) on the dependent variables (here, leaf and seed performances) are investigated and the best levels of the independent variables are found by comparing the resulting average values. However, such results may be inaccurate. The aim of the present investigation is to find accurate optimal values of nitrogen fertilizer and planting density via interpolation (optimization), so as to enhance leaf and seed performances of woad, using artificial neural network (as a complementary analysis). In a farming test with split plot design at the research farm of Jiroft Branch of Islamic Azad University (Jiroft, Iran), the effects of four planting densities (primary factor, 10, 15, 25, and 35 plants per sq. meter) and four levels of nitrogen fertilizer (secondary factor, 50, 100, 150, and 200 kg per hectare) on the weights of leaf and seed of woad were investigated. The obtained results from the analysis of the data indicated that the neural network structure based on 4 neurons was the most appropriate structure. Optimal levels of planting density and nitrogen fertilizer were found to be 32 plants per sq. meter and about 70 kg of nitrogen fertilizer per hectare, respectively, which could increase the leaf and seed performances by 6.3% and 7.7%, respectively.