A method for predicting the remaining life of marine bearings based on transfer learning and multiple time windows

Abstract

The invention discloses a method for predicting the remaining life of a marine bearing based on migration learning and multi-time windows, belonging to the technical field of wear detection technology of mechanical parts. The method includes: training a CNN aging model and a multi-time window prediction model based on an LSTM neural network; The vibration signal of the predicted bearing is input into the CNN aging model to determine whether the bearing has an early failure; if an early failure occurs, the CNN depth features of various preset length windows corresponding to the vibration signal are input into the multi-time window prediction model, and various presets are obtained. The life prediction value corresponding to each length window; fuse multiple life prediction values ​​to obtain the target predicted life of the bearing to be predicted; the invention adopts the method of fusion of multiple preset length windows to solve the problem that a single window is difficult to adapt to multiple degradation modes, The multi-time window prediction model based on LSTM neural network estimates the remaining service life of the bearing to be predicted, and the prediction accuracy is high.

Description

technical field [0001] The invention belongs to the technical field of wear detection of mechanical parts, and more particularly relates to a method for predicting the remaining life of a marine bearing based on transfer learning and multiple time windows. Background technique [0002] Rolling bearings are one of the key components in mechanical equipment, and their working conditions are directly related to the overall performance of mechanical equipment. Predicting the remaining life of rolling bearings is helpful to arrange the maintenance of equipment in advance and avoid failures and failures caused by bearing failure, which is of great significance to ensure the reliability and safety of mechanical equipment performance. The full life cycle of rolling bearings records the gradual evolution process of rolling bearings from normal state through different degrees of degradation state to final failure. Here, the degraded state refers to the stage from "early failure point...

AI Technical Summary

Problems solved by technology

[0005] In view of the above defects or improvement needs of the prior art, the present invention provides a method for predicting the remaining life of marine bearings based on transfer learning and multi-time windows, the purpose of which is to judge whether the bearing has an early failure based on the CNN aging model to judge the vibration signal. After confirming the occurrence of early failure, the remaining life prediction is performed; the adaptability of the LSTM_CORAL prediction model between different degradation models is enhanced through transfer learning, and a multi-time window fusion method is used to solve the problem that a single window is difficult to adapt to multiple degradation modes. ; thus solve the technical problem of inaccurate prediction of the remaining life of the bearing in the prior art

Images