This article uses Dynamic Bayesian Network (DBN) model, and Recurrent Neural Network (RNN) model to conduct dynamic risk assessment of the coal slurry preparation system in the coal gasification process. Firstly, based on the coal slurry preparation system's technological process, the fault tree model is further transformed into the DBN model. Secondly, the prior parameters of DBN model are obtained by Monte Carlo simulation combined with Bayesian estimation method, and the parameter learning is realized. Then, the prior parameters are substituted into the recursive reasoning algorithm of DBN model, and the Matlab language realize the bi-directional inference function. Meanwhile, validating with software inference results. Then, the RNN is combined with the DBN. To increase the forecast precision and generalization competence of DBN model. Finally, bi-directional inference is carried out on the optimized DBN model. Through analysis, weak links in the system can be determined and improvement measures can be proposed.
LIU et al. (Thu,) studied this question.