Fuzzy Traffic Signal Control. Principles and Applications

摘要

The FUSICO (Fuzzy Signal Control)-research project was started in 1996 at the Helsinki University of Technology. The main goals of the project are theoretical analysis of fuzzy traffic signal control, generalized fuzzy rules using linguistic variables, validation of fuzzy control principles, calibration of membership functions, and development of a fuzzy adaptive signal controller. This thesis discusses four hypotheses for fuzzy traffic signal control. They are (I) generality of fuzzy control, (II) competitiveness of fuzzy control (III) multilevelity, -dimensionality and -objectiveness and (IV) realisticity in real traffic signal control. The control principles and rules for the fuzzy control are modeled based on the actions of an experienced policeman represented by knowledge of an experienced signal control planner. According to the results the control parameters of fuzzy traffic signal control can be divided into three different groups: traffic volume, capacity, and level of service parameters. The fuzzy control algorithm of isolated traffic signal control can be derived based on these parameters. The fuzzy inference is perhaps the most important part of fuzzy control, but also the methods of fuzzification and defuzzification have to be introduced. Some artificial methods, like genetic algorithms and neural networks, have been tested without any benefits in comparison with the empirical membership functions. The fuzzy similarity method, which is based on Lukasiewicz's logic, has been introduced as a potential defuzzification method.