A new safety assessment model for complex system based on the conditional generalized minimum variance and the belief rule base

The safety assessment of complex system is important for implementing and fulfilling the policy of “safety first and prevention oriented”. Most available approaches cannot combine historical data with expert knowledge or cannot handle vague and uncertain information efficiently. In this paper, a new safety assessment model for complex system based on the conditional generalized minimum variance (CGMV) and the belief rule base (BRB) is proposed. In the proposed model, to decrease the computation and improve the accuracy, the conditional generalized minimum variance is used to select the key features. Meanwhile, BRB is utilized to deal with both quantitative and qualitative information under uncertainty. Moreover, to improve the precision and efficiency of BRB, the referenced values for the antecedent attributes are optimized by the fuzzy subtractive clustering algorithm. Meanwhile, the belief degrees are calculated by the modified fuzzy c-means clustering. What's more, the differential evolution (DE) algorithm is used to identify the optimal BRB parameters. The new proposed model is applied to an actual engineering system, which is used to testify the validity of the new model. Compared with other approaches, the proposed model has shown superior accuracy and less computation complexity.