Аннотации:
© by University of Guilan, Printed in I.R. Iran. The environmental impact of transport is significant because transport is a significant user of energy, and burns most of the world's petroleum. This issue creates air pollution, including nitrous oxides and particulates, and is a substantial contributor to global warming through emission of carbon dioxide. This article analyzes the Issue of Transport and the Environment, then solves the evaluation problem of the functional state of vehicle drivers based on the formation and use of a fuzzy knowledge base. The provided the classification of human functional state types. The expediency of using pupillometry as an objective method to analyze the pupillary reaction of a human eye to illumination change is pointed out to assess its functional state. The Analysis of the neural network approach is carried out to determine the functional state of a person's intoxication. It points out its main drawback associated with the impossibility of interpreting the solution obtained using a neural network. To eliminate this drawback and improve the efficiency of decision support to assess the functional state of vehicle drivers, it is proposed to use the mathematical apparatus of fuzzy neural networks to form fuzzy knowledge bases and provide their use in inference mechanisms. In this case, the solution to the problem will be a binary answer ("drunk", "not drunk") with the interpretation of the solution obtained in the form of a set of fuzzy rules written in a natural language understandable to humans. The tasks are set for the formation of a knowledge base to assess the functional state of drivers. The scheme of pupillogram initial data collection is described, as well as the stages of their preparation for Analysis. Pupillogram parameters that significantly characterize the pupillary response of a person to illumination change were identified by an expert method using the methods of correlation analysis: The minimum diameter of the pupil, the diameter of its half constriction, the amplitude of constriction and the time of half expansion. The structure of the generated data sample with the volume of 1000 records is described. A knowledge base was formed after their Analysis, consisting of 2632 fuzzy production rules. To assess the accuracy of determining the functional state of a person based on the knowledge base, a balanced test sample of 400 records (200 records of each class of functional state) was compiled. The test results showed that the number of type 1 errors was 1%, and the number of type 2 errors was 3%. The overall accuracy of determining the functional state of a person based on the generated knowledge base was 96%. The generated fuzzy knowledge base can be effectively used in decision support systems to assess the functional state of vehicle drivers when they undergo a pre-trip medical examination.