Decentralized Model-Free Adaptive Control for Signalized Intersections Network

Abstract Signalized intersections network consists of signalized intersections, which is characterized by large scale level. Decentralized control scheme is appropriate to deal with such systems, whose design and application are simpler and more practical than centralized controllers. The same as other large scale systems, it is difficult to get an accurate model of signalized intersections network; meanwhile, it has its own characteristics. The prominent one is that interactions between intersections (subsystems) are easily measured. Considering the two characteristics, we propose a new decentralized control method based on MFAC scheme. Thanks to MFAC scheme, the mechanism model is not needed. Because interactions are easily measured, our method views interactions as measurable disturbances. This just partially uses information of interactions because we do not attempt to force interactions to be some desired values. Using PARAMICS, the simulation study verifies the effectiveness of our control method.