A Temporal Bayesian Network based on uncertain time interval

In the real world, the start time and the end time of many events have uncertainty and can not be accurately represented as time interval. In order to express the uncertainty of time interval uncertainty and its state, the traditional time interval is transformed into an uncertain time interval, and a Temporal Bayesian Network based on uncertain time interval is proposed in this paper. Different from traditional Bayesian Network, such Temporal Bayesian Network takes precedence relationship into account besides causal relationship. The corresponding constructing procedure and reasoning algorithm are also given. Furthermore, the Temporal Bayesian Network based on uncertain time interval is applied to the risk analysis of dam overtopped. The results show that the proposed Temporal Bayesian Network has a better ability in expressing two kinds of uncertainty at the same time: the time interval uncertainty and the state uncertainty. It also demonstrates that the uncertainty of start time and end time has an obvious effect on the posterior probability which directly influences the decision-making.