Evidential Reasoning Using the Fuzzy Extension of Dempster-Shafer Theory: Application to the Quality Assessment of Drinking Water

摘要

This paper introduces a novel decision support system using an evidential reasoning algorithm for dealing with highly uncertain and incomplete information. The proposed algorithm provides a generic tool for modelling expert knowledge on multi-expert multi-attribute decision making problems which can arise in different areas of civil engineering including infrastructure condition monitoring, safety analysis, vulnerability assessment and risk-based environmental analysis problems. The proposed method fuses different forms of uncertain information including vague, non-specific, conflicting and partially credible judgments of experts. An evidential reasoning model based on the fuzzy extension of Dempster-Shafer theory (FDST) is presented to handle both probabilistic and fuzzy uncertainties. As an application, the proposed approach is used to assess water quality using uncertain information on multiple water quality parameters (WQP) in a distribution network. This case well represents the situation where aforementioned uncertainties exist in a utility-based environmental problem. A set of relevant microbial and physicochemical WQPs available for a distribution network and the uncertain knowledge of the expert on the water quality are considered. The results indicate that the proposed structure can successfully aggregate different bodies of evidence on water quality when various uncertainty patterns are involved.