A Fuzzy-Linear Programming Approach for Estimating Turning Movement Volumes at Intersections

Abstract

Estimating intersection turning volumes from total link volumes has been an active research area for so many decades. There is little doubt that the development of a robust approach to solve the problem can lead to significant labor and cost savings. This paper describes a new approach to estimate intersection turning counts using fuzzy membership functions and linear programming. The proposed approach requires two pieces of information, a historical seed that represents counts of a previous period and total link in- and out-flows. The approach was tested on a number of intersections and the results were assessed via error measurements. The results were benchmarked by comparing the estimating accuracy of the proposed method to the widely-used Furness model. It was found that in general, the proposed approach performed very well whenever a good seed was used in the estimation. Furness method was shown to slightly outperform the proposed Fuzzy-linear (FL) model in terms of reducing the estimation error for most case studies. In general, the estimation errors varied from one intersection to another according to volume variability. Estimation errors at the intersections with lower variability were relatively small and vice versa. The proposed approach was deemed valid and comparable to Furness model.