Genetic fuzzy logic approach to local ramp metering control using microscopic traffic simulation

Ramp metering is one of the most common solutions for improving motorway traffic flows. Fuzzy logic based ramp metering algorithms are considered to be an effective control measure to handle a complex nonlinear traffic system. In this paper, a genetic fuzzy approach is proposed to design a traffic-responsive ramp control algorithm for an isolated on-ramp. The problem is formulated as the inflow optimization of the on-ramp, based on the evaluation of motorway traffic condition. The inflow of the on-ramp is considered as the decision variable, where the ramp control problem will be treated as a nonlinear optimization problem of maximizing the evaluation function. In this paper, a well-known existent fuzzy logic control (FLC) ramp metering algorithm is implemented with an on-ramp congestion model in a stochastic microscopic traffic simulator, Aimsun. Genetic algorithm (GA) is applied to improve the performance of the fuzzy control system. The GA tunes the parameterized membership function of each fuzzy input to maintain the flow density of motorway below the estimated congestion density. The performances of the genetic fuzzy logic control (GFLC) ramp metering are compared with FLC ramp metering with respect to the percentage change of TTT (Total Travel Time). The simulation results show that the GFLC ramp metering provides significant improvement of TTT and better ability to maintain system flow density than the FLC ramp metering.