Lógica Fuzzi na recomendação de cultivares de feijoeiro comum quanto à adaptabilidade e estabilidade

Abstract

The plant breeding is a tool for higher plants to pre-existing. Soon, use tools that assist in making assertive decisions are necessary to the success of breeding programs. In this context, arises the fuzzy logic as a computational intelligence technique with wide applicability in the genetic improvement, especially in the automation of decision making in adaptability and stability studies. Thus, it aimed to employ fuzzy logic adaptability methodologies and stability in order to automate the decision-making on the recommendation of cultivars. In the first study aimed to evaluate the efficiency of fuzzy logic methodology applied by Carneiro (2015), as a tool to support the recommendation of common bean cultivars to adaptability and stability and to compare the results obtained using different methodologies presented by this author. The tests consisted of 18 common bean cultivars conducted in the cities of Ponta Grossa - PR, Santo Antônio de Goiás - GO and Uberlândia - MG, in crops of "water", "dry" and "winter" in 2006, 2007, 2008 and 2010. The fuzzy controllers were developed based on fuzzy inference system proposed by Mamdani based on the method Eberhart and Russell (1966) or in conjunction with the method of Lin and Binns (1988) modified by Ram (1998). For the analysis we used the R software through programming. The fuzzy logic methods enabled a classification of genotypes appropriately according to the parameters of each method, and showed similar results for most cases. The hybrid model, based on the methods of Eberhart and Russell (1966) and Lin and Binns (1988) modified by Carneiro (1998) allowed better distinguish the reaction of the front genotypes to environmental variations. In the second study aimed to develop fuzzy controllers to automate the decision-making adaptability studies and stability by methods Annicchiarico (1992) and Cruz, Torres and Vencovsky (1989) and check its efficiency by using experimental data with cultivars bean commonplace. Fuzzy controllers have been developed based on the Mamdani inference system proposed by the two methods of adaptability and stability studies. To check the performance of the drivers were considered test data 18 common bean cultivars grown in 11 environments. The controllers have been developed from established routines in the program R. The fuzzy controllers based on the methods of Cross and Vencovsky Torres (1989) and Annicchiarico (1992) classified 18 cultivars appropriately according to the criteria of each methods. So, fuzzy logic is a very useful tool in the automation of decision making in adaptability and stability studies, it is important for genetic improvement.