Dynamic multi-level system modeling and state prediction method based on hybrid cognition

Abstract

The invention discloses a dynamic multi-level system modeling and state prediction method based on hybrid cognition. The method comprises the following steps: step 1, performing system analysis by using a fault tree and a static Bayesian network method; 2, analyzing the dynamic multi-level system by a hybrid cognitive method, and constructing a static Bayesian network B = (B1, theta); 3, expandingthe static Bayesian network into a dynamic Bayesian network to form a system state prediction model; 4, reasoning, evaluating and predicting the health state of the dynamic multi-level system. According to the hybrid cognitive system analysis method provided by the invention, the problem of incomplete cognition of a fault tree and a static Bayesian network method on a dynamic multi-level system can be solved, and the prediction accuracy of a static Bayesian network model established on the basis of hybrid cognition on the system state is higher. According to the system state prediction modelestablished by the invention, the overall state of the system can be reasoned, evaluated and predicted by utilizing the basic-level component data, and the state change trend of the system is mastered.

Description

technical field [0001] The present invention is aimed at the data of basic-level components in a dynamic multi-level system. A hybrid cognition method is formed by combining a fault tree with a static Bayesian network and a STAMP method to analyze a dynamic multi-level system. A dynamic Bayesian network constructs a system prediction model, which belongs to the technical field of system modeling and state prediction. Background technique [0002] For dynamic multi-level system products, it is very necessary to carry out state prediction modeling for them. By building a system prediction model, the state of the system can be evaluated, the state of the system can be mastered, and accidents can be avoided. Can guide follow-up system maintenance and other work. At present, for dynamic multi-level systems, the state prediction modeling is mainly realized by classical methods such as fault tree analysis, static Bayesian network, and the combination of fault tree and static Bayes...

AI Technical Summary

Problems solved by technology

However, in the study of classical analysis methods, the model obtained by fault tree analysis is often directly converted into a static Bayesian network structure, so that the structure of the established model depends entirely on the results of fault tree analysis and cannot find hidden relationships in the network. At the same time, the established model cannot continuously predict the state of the system over time, resulting in a large deviation between the evaluation results of the constructed prediction model and the actual state in practical applications

Images