Methods for Selecting Linguistic Variables in the Fuzzy Traffi c Light Control System

To increase the capacity of the intersection and simultaneously reduce the travel time of the vehicle, optimization of traffic light control is necessary. The existing traffic light control systems cannot control dynamic systems in which several factors influence the decision-making process. The determination of factors (output variables) and the fuzzification process are the main problem of the fuzzy logic algorithm, and the quality of the compilation of the term set of input linguisticvariables and the definition of the function of belonging affect the optimal control of the light signals. The article provides an analytical overview of the ways of using linguistic variables for fuzzy inference systems when controlling traffic light signals. The subject of the article is the input linguistic variables for decision-making in a fuzzy management model. The analysis of modern research is presented and the main input linguistic variables are described. In the first section of the work, the general principle of building a rule base for fuzzy inference systems based on the Mamdani and Takagi-Sugeno methods is considered. The following sections are devoted to the peculiarities of such output linguistic variables that affect the operation of a fuzzy traffic light, such as: the number of vehicles, the current time of the green signal, road users (pedestrians), weather conditions and the number of lanes (width) of intersected roads. Accounting for these variables, their fuzzification and the formation of an appropriate rule base for the design of fuzzy systems is a very difficult task. In this regard, one of the key problems is precisely the problem of choosing the necessary input parameters depending on the type of intersection.A review of the literature has shown that the research of the fuzzy controller in traffic management is still at the initial stage of development. Many of the unresolved issues raised in ozor can be addressed in further research