Traffic Signal Control Using Genetic Decomposed Fuzzy Systems

In this paper, Decomposed Fuzzy Systems (DFS) structure is applied to design single intersection signal fuzzy controller. The DFS structure is to decompose each fuzzy variable into layers of fuzzy systems and each layer is to characterize one traditional fuzzy set. DFS adjusts the fuzzy membership function, the leading part, and enriches the fuzzy rule base through structural changes, thus provides the system with more adjustable parameters to facilitate possible adaptation in fuzzy rules, but without introducing a learning burden. It also can be found that the function approximation capability of the DFS is much better than that of the traditional fuzzy systems. At the same time, in order to solve possible defects brought by expert experience, Genetic Algorithm (GA) is applied to the optimization of DFS rule base in this paper. Taking the four-phase single intersection as a case study, an intersection signal control algorithm is obtained using the proposed DFS based on Genetic Algorithm (G-DFS). Simulation results show that the G-DFS controller reduces average vehicle delay, queuing length, average parking rate, and average vehicle travel time effectively, and the controller can smoothly adapt to different traffic flow changes.