Use of Fuzzy Sets to Evaluate Driver Perception of Variable Message Signs

Abstract

Current methods for evaluating the quality of service provided by a variable message sign (VMS) may not yield results that represent the satisfaction drivers experience because these techniques cannot represent the variability and complexity of human perception with great fidelity. To solve those problems, a new method that applies fuzzy set theory was developed. Results of a preexisting survey of VMS service quality were reanalyzed with this method. For this application, two membership functions were constructed with two different experimental methods: interval estimation and pairwise comparison. Specifically, construction of the second membership function used Saaty's eigenvector method. These two membership functions employ five linguistic statements to represent the degree of satisfaction and relative importance of six performance criteria. Quality of VMS service perceived by an individual driver was evaluated with the concept of the fuzzy weighted average. A set of 322 quality measures was computed, and they were aggregated and transformed into a single percentage value with the use of an arithmetic fuzzy mean. The defuzzified final value indicates the degree of satisfaction with VMS service perceived by a group of participating drivers, taking into consideration the variance of human perception and the degree of importance of the six criteria. By employing the proposed method, the quality of VMS service as subjectively perceived by humans and influenced by various exogenous factors can be evaluated while still considering the subjective and complicated manner of human thinking.