A model for remaining useful life interval prediction of servo turret power head system of turn-milling center under time-varying operating conditions

With the diversification of machining tasks in turn-milling centers, the service conditions of the servo turret power head system are complex and changeable, and there are multi-source uncertainties in the degradation monitoring process. Based on the improved conditionally parameterized convolutions and nonlinear Wiener process, this paper proposes an interval prediction model suitable for the remaining useful life (RUL) under time-varying operating conditions. Firstly, a method for making sample performance degradation labels based on operating condition classification is proposed, and the labels under continuously identical operating conditions are linearized according to the classification results of operating conditions to solve the problem of inconsistent degradation rate under time-varying operating conditions. Then, a conditionally parameterized convolutions module considering global–local features (GL-CondConv) is proposed, and the convolution kernel parameters are adaptively learned according to the input samples, so that the model fully considers the influence of the features of each sample on the prediction results under time-varying operating conditions. Finally, the nonlinear Wiener process is used to estimate the RUL interval of the equipment to quantify the RUL uncertainty. The effectiveness of the proposed method is verified on the servo turret power head system dataset and PHM bearing dataset.