Application of Fuzzy Inference System to Estimate Perceived LOS Criteria of Urban Road Segments

Abstract

Perception based Level of service methodology depicts how well a transportation facility satisfy its road users. There are several engineering factors affecting driver’s perception of service quality on urban arterials, which vary largely by individual user and not easy to be distinguished. Therefore, user-perceived service quality appraisals don’t relate to the LOS assessed using Highway Capacity Manual (HCM) method. The main objective of this study is to speak to the variability of human discernment and to characterize level of satisfaction of road users using hierarchical fuzzy inference system (HFIS). Around 500 responses of road users were gathered from two emerging cities of India i.e. Bhubaneswar and Vishakhapatnam, where traffic facility is portrayed by distinctive sorts of roadway features and variation in number of pedestrians, motorists and each category of vehicles. So that our survey can include every circle of transportation and people belonging to all classes can be taken into account. The present approach, contains fuzzy reasoning experiences to combine important factors which reflect road user’s perception of service quality. In order to decrease the intricacy of multivariable, the fuzzy inference system (FIS) is partitioned into two subsystems, together conducing to a hierarchical fuzzy inference system. First the four input variables namely roadway geometry, traffic facilities, pavement condition and safety are fuzzified. The resulting fuzzy values are entered into the fuzzy inference system, where number of fuzzy rules are generated by applying fuzzy operators. Applying max–min composition method all the outputs are aggregated and defuzzified to evaluate composite LOS of various road segments of both the cities. This methodology offers unique and significant approach to define service quality provided by the transportation infrastructure and may defeat the constraints of customary delay-based methods to some extent.