On the Application of Bayesian Network for Dam Risk Assessment

摘要

Bayesian network (BN) provides a robust probabilistic method of reasoning under uncertainty. ThernBN has been developed and applied mostly in the filed of artificial intelligence. BN has a large potential for thernapplication in the modeling of natural hazards and the corresponding risks. Recently, BN has been successfullyrnapplied in risk assessment for technical systems. However, the author is aware of only few reported applicationrnof BN in the filed of dam risk assessment. The objective of this paper is to demonstrate the potential andrnadvantages of BN for the application in dam risk assessment. For this purpose, a general framework for damrnrisk assessment is presented and a basic introduction to BN is provided. The application of the BN is illustratedrnconsidering the problem of risk mitigation due to existing dam failure with a limited, reasonable expenditure ofrnfunds, where it is shown how Bayesian networks can model the dam risk assessment process effectively. Thernresults indicate that Bayesian network has a large potential for the modeling of dam risk assessment because ofrntheir intuitive format, which facilitates the cooperation of specialists from several disciplines. The methodologyrncan aid in the identification of which dams are candidates for modification and in selecting the appropriate typernand level of modification to be made.