Development a Risk Assessment Model Using Weight Fuzzy Bayesian Theory

• Main Authors: Chi-Feng Wu, 吳琪楓
• Other Authors: Shuen-Tai Ung
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/07674915585302770963

碩士 === 國立臺灣海洋大學 === 商船學系所 === 99 === 　　Risk assessment is the process of assessing and combining the variables of events to obtain an index expressing the degree of risks. Its purpose is to reduce risks to an acceptable level to eliminate or reduce the damages caused to human beings and properties. However, since maritime accidents cannot be recovered as fast as those occurring in land, a proact risk assessment that is capable of providing risk results beforehand is important. It is particularly true for maritime risk assessment where the lack or incompleteness of data exits and the development of effective risk methods is therefore crucial. When evaluating risks, it is very often that the relative importance of each variable may be different. Traditional methods, however, ignores such situations.. Some other approaches, on the other hand, consider the weight of each variable, the process of which are nevertheless, regarded as complex and time consuming. In addition, the membership functions of linguistic terms in the consequent part of rules are often not considered. This will lead the risk results not able to be genuinely acquired and may be distorted. Accordingly, the objective of this research is to establish a methodology that is capable of solving the difficulties aforementioned and providing reasonable outcomes under circumstances where the lack of data exists. 　　The framework is constructed based on the Fuzzy-Bayesian theory. By virtue of the establishment of fuzzy rule-base, the relative weight of each factor and the threshold of the consequence part of rule can be considered. The opinions with regard to each level of the linguistic terms in the consequent part of rules are also considered. Therefore, the results generated from experts can be genuinely acquired.. In addition, through the use of the belief degree approach to the consequent parts of each rule and the applications of the Fuzzy Delphi method and Bayesian theory, the evaluation process is simplified. It can be further facilitated by the introduction of risk software, Hugin, to generate probabilistic risk results. The methodology is validated by a maritime case study selected from the previous research. It is concluded that the framework proposed is capable of providing reasonable outcomes when the weights of variables are various without distorting the expert opinions.