Product key part life end-to-end prediction method based on self-attention network

Abstract

The invention discloses a product key part life end-to-end prediction method based on a self-attention network. The method comprises steps of installing acceleration sensor on a key part of a mechanical product, recording vibration signal data of the part through the acceleration sensor, and recording the operation time of the part; performing data preprocessing on the vibration signal data, and processing the operation time at the same time to obtain a training data sample and a corresponding remaining service life label; constructing a prediction model of the remaining service life; training the prediction model to obtain a trained prediction model; and collecting vibration signal data of the part, performing data preprocessing, inputting the data into the trained prediction model, and performing real-time prediction to obtain the remaining service life of the part. The self-attention network is adopted to construct the prediction model, time sequence features in the vibration signals can be fully mined, the defects of complex data preprocessing and feature extraction are overcome, end-to-end prediction from the vibration signals to the remaining service life is achieved, and the method has the advantages of being easy to operate and high in generalization.

Description

technical field [0001] The invention belongs to an end-to-end prediction method for the life of a key product part in the field of prediction of the remaining service life of a mechanical product part, and relates to an end-to-end prediction method for the life of a key product part based on a self-attention network. Background technique [0002] Key parts of mechanical products such as knives, gears, bearings, etc. are widely used in modern industry. They not only play a key role in the realization of the functions of mechanical products, but also have a great impact on whether the mechanical products can work normally. Taking bearings and gears as an example, if the bearings and gears of the reducer fail, the reducer will not be able to function normally. Remaining service life prediction technology is an important part of product health management. It can be used to monitor the operating status of key parts of the product in real time and provide the time when they can st...

AI Technical Summary

Problems solved by technology

However, the existing methods still have certain limitations, such as the model’s weak time-series feature extraction ability, which makes it impossible to further improve the prediction accuracy

In addition, existing methods usually rely on complex data preprocessing and feature extraction to pre-extract important features in the data, which cannot achieve end-to-end prediction

In this case, the prediction accuracy of the model often depends on the features extracted manually, and inappropriate features will greatly reduce the prediction accuracy of the model

At the same time, accurate data preprocessing and feature extraction require a lot of professional knowledge in this field, and the implementation process is relatively complicated, which increases the difficulty of applying existing prediction methods in the industry

For example, in the patent "Prediction Method for Remaining Life of Key Parts of Products Based on Asymmetric Loss Neural Network", before the deep learning model processes the data, it is necessary to extract the features in the vibration signal through statistical means and wavelet changes, which limits the The generalizability and feasibility of the method

Images