Dynamic optimization of watering for maximizing the sugar content and size of Satsuma mandarin using intelligent approaches

Abstract Larger size and higher sugar content fruits are desired for fruit production. However, these factors compete with each other. For example, water stress (smaller watering) causes an increase in sugar content but a decrease in fruit size. Therefore, there exists an optimal watering operation for their maximization. In this study, an optimal watering scheduling that maximize both of the sugar content and the diameter of Satsuma mandarins grown in the field was investigated using neural networks and genetic algorithms. The monthly data of the fruit responses and climate factors were measured for identification from August to November, 1996-2009. Dynamic changes in the sugar content and diameter of the Satsuma mandarins, as affected by the rainfall (watering), sunshine duration and air temperature, were first identified using neural networks, and then an optimal watering scheduling (rainfall management) that maximizes the sugar content and size of the Satsuma mandarins was determined through simulation of the identified neural-network model using genetic algorithms. The optimal value obtained here was a combination of the marked increase in watering during the fruit-developmental stage (August and September) and a significant decrease in watering during the fruit-maturing stage (October and November). From the model simulation, a marked increase in watering during the former stage induced an active developmental growth of the fruit, and a significant decrease in watering during the latter stage induced an increase in the sugar content. Drip irrigation is commonly used for increasing the watering whereas plastic-film mulching is used for reducing it.