Learning failure prediction bayesian networks based on genetic algorithms

To establish practical models and facilitate engineering applications, this paper proposes a novel learning algorithm for failure prediction Bayesian Network (FPBN) modeling. At first, the basic concept of FPBN is presented, including its node class, edge orientation and conditional probability distributions. Then, the corresponding learning algorithm of PFBN is developed by applying the advantages of genetic algorithm and the operation steps of this algorithm are described in detail. At last, the simulation study is implemented based on the generated helicopter convertor operation datasets. The comparison results of different FPBN learning algorithms show that the proposed method can build the objective model well with least evolution generations.