Ensemble Influence Nets for Equipment Health Status Classification

As a simplified model of Bayesian Network, Influence Nets (IN) has a wide range of applications in the description and modeling of uncertain causality. However, too many nodes in IN will increase the number of parameters in modeling and limit the applicability of IN model to high-dimensional problems. In this paper, based on the bagging framework, we propose an Ensemble-IN model to overcome this flaw. Firstly, several weak IN models are integrated coherently, each of which only consists of a subset of nodes. Then, difference evaluation algorithm (DE) is introduced to optimize the parameters in weak IN models. Finally, the combination methods is utilized to integrate the analysis results of weak IN models. The rationality and feasibility of the proposed Ensemble-IN model are verified by a practical case of health status classification of engine.