Sistemas inteligentes aplicados na modelagem da produção de hastes florais de Heliconia spp

Abstract

The objective of this work was to analyze the influence of photoperiod (N, hours), air temperature (TM, °C) and precipitation (PREC, mm) in the production of flower stem of four species of Heliconia spp., evaluate the temporal stability of production and develop models to estimate the production of flower stem by Artificial Neural Networks (ANNs) and multiple regression. The database used in this study came from an experiment conducted in the period of dec/2003 until mar/2012, in Camaragibe-PE. Four cultivars of heliconias were evaluated: H. psittacorum L.f. Red Opal, H. psittacorum x H. spathocircinata cv. Golden Torch, H. stricta Fire Bird e H. rauliniana. We obtained the number of flower stem harvested per plant per month (HCM) and weekly (HCS), TM, PREC and N. To analyze the temporal stability was known the relative difference parameter (HCM) each month. Overall species showed stability in the production of HCM, which we determined the months of lower and higher production. The largest increases in HCS were observed in cv. Golden Torch showed that in the month of December a HCM 87% higher than the average annual and H. rauliniana with HCM 274% higher for the month of November. The use of ANNs compared to multiple regression was more efficient for predicting the production of flower stalks of four species of Heliconia spp. based on TM, N, and PREC. For all species studied the results of the simulations of increased air temperature indicated a fall in production which ranged from 33.19% to 38.71%.