A Health State Prediction Model for Aeroengine Based on Multi-attribute Belief Rule Base with Considering Monitoring Error

The aeroengine is a type of complex mechatronic system that has significant nonlinearity and uncertainty, and it is difficult to accurately establish the mathematical model for the health status prediction of the aeroengine. Fuzzy theory can effectively use the fuzzy information and knowledge provided by experts to predict the health state of the aeroengine without a precise mathematical model. Multi-Attribute Belief Rule Base (MBRB) is an excellent multi-attribute decision-making tool for solving nonlinear, uncertain problems in the aeroengine by introducing expert knowledge and fuzzy theory. However, the monitoring error of multi-attribute data greatly affects the decision-making accuracy of MBRB. The quality of monitoring data is greatly influenced by environmental interference and sensor quality. Therefore, the distance-based and variance-based methods are used to quantify the monitoring error of multi-attribute data, and a new method for calculating the degree of match is proposed for merging into the process of ER reasoning. Then, a health state prediction model for aeroengine based on a multi-attribute belief rule with considering monitoring error (MBRB- \({\varepsilon }\) $\epsilon$ ) is proposed in this paper. The initial subjective parameters that may not be suitable for engineering practice are given by experts in the model. The projection covariance matrix adaptive evolutionary strategy (P-CMA-ES) is used as the optimization algorithm for training initial parameters. The results achieved are investigated by two aeroengine data sets to verify the effectiveness of the model. The results show that the health state prediction model of an aeroengine based on the MBRB- \({\varepsilon }\) $\epsilon$ model can accurately predict the health states of the system.