Life Control Based on Limited Incremental learning and WE-OSELM

This paper coordinates the differences between the connotations of remaining life control and reference life control, and adopts a data-driven life prediction method to carry out real-time differential life prediction under unknown failure mechanism conditions. Based on the research of WE-OSELM algorithm, this paper integrates the design of multi-step prediction method and proposes a remaining life prediction method and prediction model based on finite incremental learning and WE-OSELM. Firstly, in view of the fact that the number of hidden layer nodes affects the performance of the prediction algorithm, the finite incremental learning method is used to quickly screen the optimal number of hidden layer nodes within a limited range, and the IPW-OSELM algorithm training is quickly completed by integrating historical degradation data and online time series data. Then, the error data in the training process of the IPW-OSELM algorithm is used to train the error compensation algorithm. Finally, the failure threshold and multi-step prediction results are combined to determine the remaining life of the prediction object. Simulation comparison shows that the algorithm in this paper not only has a high conservative remaining life prediction ability, but also has the ability to accurately predict to a certain extent, and the results are highly credible.