Studying the importance of fuzzy sets definitions for fuzzy signal controllers

The fuzzy sets adopted for the development of fuzzy signal controllers, and their respective membership functions, are recognised as important aspects related to controller operations. However, the impact of their definitions on traffic signal performance has not been sufficiently explored in the literature. Our aim is to contribute to overcoming this problem. It presents an analysis of the implications of small modifications to some fuzzy sets' parameters for fuzzy signal controllers. This analysis focuses upon three aspects. The first looks at modifications to fuzzy signal control tables, generated through MATLAB software and due to variations in the input variables' fuzzy sets. The second is the impact of the use of these different control tables on some controller operation output characteristics, as is the green time provided for each controlled intersection approach and cycle length. The last aspect is related to differences observed in traffic performance measures caused by the modifications to the controller's operation characteristics. In order to carry out the analysis of the last two aspects, a simulation study was conducted through HUTSIM simulation software, developed at the Helsinki University of Technology. The study's results show that, in many of the aspects considered, the impact of small modifications to the fuzzy sets' parameters is statistically significant at the risk level of 5%.