Operational reliability prediction of equipment based on relevance vector machine

Due to lack of failure data, traditional statistics-based reliability evaluation methods cannot effectively evaluate the reliability of mechanical equipment. The operational reliability concept was introduced in this work to describe the running condition of the equipment, and a prediction method for the operational reliability based on relevance vector machine (RVM) was proposed. In the method, the operational reliability calculation model was adopted firstly to establish the relationship between status indicators of the equipment and the operational reliability, and then the relevance vector machine was introduced to predict the trend and the corresponding probability distribution function (PDF) of the status indicators. Finally, the predicted values and PDF of status indicators were inputted into the operational reliability calculation model and the predicted operational reliability of the equipment was then obtained. The method was applied to the operational reliability evaluation of aero engine bearings, the result show that the error between the predicted failure time and the actual failure time was within 5%, which proves the rationality and effectiveness of the method.