Abstract:

The quantification of human error probability is researched. Firstly, the data resources that can be used in the quantification of human error probability are introduced, including general data, expert data, simulation data, and spot data. Their characteristics are analyzed. Secondly, the basic idea of Bayesian information fusing is analyzed. Two key problems are emphasized, which are the formation of prior distributions and the determination of fusing weights. Finally, the new method is presented, which quantifies the human error probability based on Bayesian information fusing. The first three kinds of data are regarded as prior information to form the fused prior distribution. The Bayesian method is used to synthesize all the data and get the posterior distribution. Based on the posterior distribution, the human error probability can be quantified. An example is analyzed, which shows the process of the method and proves its validity.