Conditional Value-at-Risk Optimization of Traffic Control at Isolated Intersection

We propose a new method for optimizing fixed-time traffic control plans and timings at isolated intersections by minimizing the Conditional Value-at-Risk (CVaR) of the intersection's average delay over an extended peak period. The optimization algorithm was implemented in Matlab using a scatter search algorithm to minimize the CVaR of the average delay which is based on the HCM formula for multiple period analysis. The proposed method incorporates the risk analysis and time dependent delay estimation. Numerical experiments were performed with AIMSUN in a reference intersection for an extended peak period of 3 hours to provide evidence of the advantages of this optimization method with respect to conventional optimization techniques through the analysis of 48 experiments.