An evidential similarity-based regression method for the prediction of equipment remaining useful life under complex conditions

Remaining useful life (RUL) is an important part of prognostic and health management (PHM). An evidential similarity-based regression method for the prediction of equipment remaining useful life in presence of incomplete degradation trajectories and varying operational conditions is proposed. In actual production, the working conditions of equipment will change due to the different processing workpieces, which will affect the prediction of residual service life. Therefore, the influence of different working conditions on remaining useful life (RUL) prediction is eliminated through data processing. At the same time, because of the safety factors in actual production, some key components are often replaced before failure, resulting in incomplete degradation track (i.e. the fast failure part of the equipment's data during operation is not recorded), combined with the similarity regression method and evidence theory to achieve the comprehensive utilization of the complete degradation track and the incomplete degradation track, so that the prediction method is more in line with the actual situation. On this basis, the health factor index is introduced to quantitatively reflect the health status of the equipment, further improve the prediction accuracy, and realize the prediction of the remaining useful life under the complex conditions such as the change of working conditions and incomplete degradation track in the actual production.