Neuro-fuzzy modelling of workers trip production

This paper attempts to introduce the application of Neuro fuzzy techniques for fulltime worker trip production estimations in Adelaide metropolitan area using the household/person characteristics such as age, vehicle ownership and distance from CBD. In the last 30 years, several linear regression models have been developed for this purpose. These models' linear structure does not seem suitable to predict highly nonlinear behaviour of urban transport systems. Consequently, intelligent modelling methods, as powerful nonlinear tools, have attracted much attention in the prediction of trip productions. In 1993, fuzzy logic and artificial neural networks were combined and neuro-fuzzy technique was emerged to model engineering systems. Since then this technique has been improved drastically and utilized to model a wide variety of complicated engineering systems. In this research, the aforementioned method is employed for modelling person/worker trip productions. After subtractive clustering, a meaningful relation between distance of residence from CBD area and workers' trip productions was not observed in this research. The modelling was accomplished with and without this factor and this view was justified. At the end, a fuzzy inference system was achieved which explains persons behaviour with a reasonable error range.