Comparison an artificial intelligence-based model and other models: signalized intersection delay estimates

This paper presents an adaptive neuro-fuzzy inference system (ANFIS), which has been adapted as an alternative to other classical models for estimating the vehicle delays at signalized junctions. Rules, fuzzification, and inference were modeled by ANFIS. In this model, a hybrid algorithm was used for training and tests. The artificial network used three input variables representing simulation of the time, the number of approaching vehicles in the green duration, and the number of queuing vehicles in the red duration. The results of neuro-fuzzy networks were also compared with Observation, Webster, HCM, Multiple Regression Analyses, and Signal Simulation Model (SSM) results. It was found that ANFIS, Regression, and SSM results are show the most agreement with the observation values.