Research On Signal Timing Optimization Based On Genetic Decomposed Fuzzy Systems

The transportation system is a non-linear system with randomness,complexity,and uncertainty.It is difficult to establish accurate mathematical models.Fuzzy systems can deterministically express uncertain language,thus having a good advantage in system modeling.Based on the fuzzy set theory and genetic algorithm,this thesis proposes a method for optimizing signal timing using Genetic Decomposed Fuzzy Systems.The main works are as follows:First of all,this thesis designs a decomposed fuzzy controller in a single intersection.The controller adjusts the green signal ratio in real-time according to the maximum queue numbers of the current green light phase and the next green light phase.Both input fuzzy variables and output fuzzy variables of the controller select triangular membership function,and the discourse domain is set according to actual traffic conditions.Through the decomposition of input fuzzy variables,the number of fuzzy rules increases,and the system has more adjustable parameters to capture possible changes of the antecedent part and the posterior part of rules.Simulation results show that compared with the type-1 fuzzy controller,the decomposed fuzzy controller reduces average vehicle delay,queuing length,parking rate,and average vehicle travel time by 15% to 27%,achieving the goal of optimizing signal timing at intersections.Secondly,considering that the acquisition of fuzzy rule bases in Decomposed Fuzzy Systems is still based on expert experience and manual operation,this thesis uses genetic algorithm to optimize the fuzzy rule base of Decomposed Fuzzy Systems,so that the parameters of rule base can be adjusted independently when traffic flow changes.It has been proved by experiments that the optimized decomposed fuzzy controller decreases various traffic evaluation indicators by 19% to 33% than the type-1fuzzy controller.Finally,this thesis further studies the application of Genetic Decomposed Fuzzy Systems in arterial intersections.Adjacent intersections are used as the basic control unit and the correlation between adjacent intersections needs to be considered.Therefore,this thesis designs Genetic Decomposed Fuzzy coordinated secondary control of double intersections.The first level of control is the coordinated control between intersections.The second level of control performs Genetic Decomposed Fuzzy control of a single intersection.Simulation experiments prove that Genetic Decomposed Fuzzy Systems are still effective in optimizing adjacent intersections signal timing.