THE USE OF FUZZY SETS MATHEMATICS IN PAVEMENT EVALUATION AND MANAGEMENT (UNCERTAINTY, FUZZIFICATION, PSR)
Abstract
A methodology for ranking pavement sections according to the maintenance urgency has been developed for the state of Indiana using fuzzy sets theory. Concepts of fuzzy sets theory are shown to supplement the existing procedures by methodically handling the human and system uncertainty in the pavement management system. Pavement sections in the highway network are grouped according to their maintenance needs, using the fuzzy PSI obtained from a PSR--RR (Roadmeter reading) correlation. This aids in scheduling the relevant tests for further evaluation and ranking for each pavement category. The concept of a fuzzy PSR which accounts for both the uncertainty inherent in each rating and the relative perceptiveness of each rater is introduced. Two novel approaches have been laid out for the correlation of PSR with RR. In one approach, RR is treated as a fuzzy number due to the imprecision and variability associated with it and a fuzzy relation is used for correlation. In the second approach the current notion of random Roadmeter variability is retained and correlation is done through fuzzy regression analysis. Variability inherent in the Skid-tester and the Dynaflect is also incorporated by considering the respective readings as fuzzy numbers. This is achieved by formulating a direct but an efficient fuzzification technique. It is shown how fuzzification can also be applied to methodically account for the subjectivity in the evaluation of distress manifestation of pavements. The resulting fuzzy pavement condition rating (PCR) is certainly a step towards the need of improving the condition surveys. The final ranking scheme is formulated using fuzzy multiattribute decision making concepts. The attributes relevant to each category of maintenance are identified and an expert knowledge base containing priority values for known attribute value combinations is formed in collaboration with decision makers. The ranking scheme is capable of handling possible fuzziness in the expert priorities as well. Finally it is shown how the fuzzy attribute values for each section can interact with the expert knowledge base to produce a unique set of rankings for the pavement sections.
Degree
Ph.D.
Subject Area
Civil engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.