Background: Artificial intelligence systems have been employed for the development of predictive models that estimate many agricultural processes.
Results: In present study, the predictive capabilities of artificial neural networks (ANNs) were evaluated with respect to assessing fruit firmness as a postharvest life index, with determinations made at four stages of storage: 1, 60, 120 and 180 days after harvesting. Single concentrations of nitrogen (N), potassium (K), calcium (Ca) and magnesium (Mg) on fruit (D1 ), all of these nutrient concentrations (D2 ), the ratios of the nutrient concentrations alone (D3 ), and a combination of nutrient concentrations and their ratios (D4 ), were considered.
Conclusion: The results obtained showed that fruit firmness at 1 and 60 days after harvesting was not influenced by nutrients. However, the ANN model estimated fruit firmness of 120 and 180 days, respectively, for D1 and D3 more accurately than for the D2 and D4 datasets. Application of D3 (nitrogen/calcium ratio) as the input dataset improved predictions of fruit firmness, with a correlation coefficient of 0.85 between the measured and estimated data. © 2019 Society of Chemical Industry.
Keywords: ANN; Actinidia deliciosa; calcium; firmness; shelf life.
© 2019 Society of Chemical Industry.