Equipment residual life prediction method based on double-layer attention network multi-domain feature fusion

Abstract

The invention discloses an equipment residual life prediction method based on double-layer attention network multi-domain feature fusion, and the method comprises the steps: 1, fusing domain knowledge, and extracting the time domain, frequency domain and time-frequency domain statistical features of signal information; 2, constructing a multi-scale feature extraction module, and extracting features from multi-scale signal information by adopting one-dimensional convolutional neural networks with convolution kernels of different sizes; 3, performing weighted combination on the obtained multi-scale comprehensive information based on a scale attention module; 4, inputting the multi-scale features into a long short-term memory network to extract multi-scale depth features; and 5, endowing different weights for the multi-scale depth features and the three statistical features by adopting a domain attention module, and performing equipment residual life prediction by pointedly utilizing different feature combinations. According to the method, the residual life of the equipment can be quickly and accurately predicted, so that the reliability and the safety of high-end equipment of engineering machinery are improved, and the risk of failure events is reduced.

Description

technical field [0001] The invention belongs to the technical field of equipment failure prediction and health management, and in particular relates to a method for predicting the remaining life of equipment based on multi-domain feature fusion of a two-layer attention network. Background technique [0002] With the development of science and technology, modern equipment has become more precise, automatic and systematic. In modern manufacturing scenarios, the working environment of equipment is often harsh and changeable. Long-term operation will lead to a decrease in the reliability of the equipment, and if the operating capacity of the equipment drops to a certain threshold, it may cause the equipment to suddenly shut down, which may lead to serious casualties. Therefore, the remaining useful life prediction of equipment has been relied upon in both academia and industry. Recently, data-driven residual life prediction methods do not rely on any failure mechanism, which c...

AI Technical Summary

Problems solved by technology

Most studies develop several deep learning methods, such as improving the network structure to improve estimation performance, without considering the advantage of taking advantage of traditional statistical features

[0005] (2) The importance of the scale of information in feature extraction is not fully considered

However, using too short or too long time scale for feature extraction may miss important information, and it is impossible to comprehensively find the deep correlation between time points in the mechanical vibration signal data to predict the remaining life of the equipment

Images