Different methods for the recognition from patterns exist, in article appears the recognition of patterns through the use of fuzzy membership functions, elaborated for the case of study, and elements of the fuzzy logic. The shown procedure contains four steps that they initiate in the data collection, the fuzzyfication, the calculation of the coincidence probability and the desfuzzyfication step that defines the patterns of failures in the analyzed motor of internal combustion. It was possible to be verified that the variables Tf (temperature of the fuel), Pf (pressure of the fuel) and Pai (pressure of the air), are altered before the sprouting of the studied failures, modifying his values and indicating different symptoms from incipient failures and that the patterns of failures, for each one of the three mentioned variables, settle down between the values limits according to the characteristics of each variable and following the behavior described by the developed functions of property.