In time-series environments, uncertain knowledge among variables in a time slice can be represented andmodeled by a Bayesian network (BN). In this paper, we are to achieve the global uncertain knowledge during a periodof time for decision-making or action selection by fussing or combining th