Neurofuzzy Traffic Signal Control

merkitttvsti pienilll liikennemmmrilll. The aim of this work was to create an adjustable fuzzy traac signal controller. An existing fuzzy traac signal controller was enhanced with a learning algorithm. This adjustable controller can modify its parameters in diierent traac situations, and thus reach a better control result. The performance of the traac signal controller is measured by the delay o f v ehicles. A fuzzy traac signal controller uses linguistic rules such as if the approaching traac volume is large and the queuing traac volume is small, then the green signal is long. The fuzzy concepts large, small and long are presented using membership functions. Neural networks consists of simple processing elements interconnected as a struc-tured network. In a neurofuzzy controller, the parameters of the fuzzy membership functions are adjusted using a neural network. The neural learning algorithm in this work is reinforcement learning. The neurofuzzy system under consideration is such that the most usual neural learning algorithms cannot be used. The adjustable traac signal controller is studied in a traac simulation system which includes a fuzzy signal controller. The neural learning algorithm is realized in a Matlab program which i n teracts with the traac simulation system. The learning algorithm is found successful at constant traac volumes. Starting from the initial membership functions, the learning algorithm modiies the parameters of the membership functions in diierent ways at diierent but constant traac volumes. The membership functions after the learning produce smaller delays than the initial membership functions. The learning algorithm is not found successful in situations where the traac volume changes rapidly. An additional contribution of this thesis is a small manual modiication in the rule base of the fuzzy traac signal controller. This modiication reduces the delays signiicantly at low traac volumes. Not borrowable till: Library code: iii Preface I warmly thank Prof. Harri Ehtamo for his supervision and the encouragement he has given me. My instructor, Lic.Tech. Jarkko Niittymmki and Prof. Matti Pursula have provided me with an interesting topic and a good working environment, for which I am most grateful. I wish to thank my colleagues at the HUT Transportation Laboratory. I am particularly indebted to Lic.Tech. Iisakki Kosonen and Mr. Mikko Lehmuskoski for helping me with the traac simulation system, and to M.Sc. Jari Kurri for reviewing and discussing my thesis. I also thank Dr. Esko Turunen for the comments he gave on my manuscript. My …