Agro-morphological characterization of three species of Vetch using artificial neural networks

Despite the diversity of species of annual forage legumes that may be grazed in Algeria, few have been used to feed livestock. The use of these forages can improve livestock nutrition in the context of sustainable development. The valuation of plant genetic resources and knowledge of species and research pastoral interest in food is of paramount importance hence the choice of this work aimed at the agro-morphological characterization of three species of vetch ((Vicia ervilia, V. narbonensis and V. sativa) in the semi-arid region of Setif - Algeria-. Since data on these variables are characterized by uncertainty, vagueness and complexity, we found it useful to analyze them with an artificial neural network system as a technique of artificial intelligence. We propose in this reading performance of the study a package rates. It is then necessary to relate the parameters species, variety and quantitative variables that the input space to the study area in terms of actual measured values. The function mapping input parameters performance is adjusted by the network from a set of real measured values of pitch. After learning all random variables introduced at the input used to read the expected yield of the output