A multi-source heterogeneous information fusion fault diagnosis method based on two-level transfer learning

Abstract

The present invention proposes a multi-source heterogeneous information fusion fault diagnosis method based on two-level transfer learning, the steps of which are as follows: firstly, using the trained VGG16 model to extract the features of the monitoring screenshot images, and taking the features of the monitoring screenshot images as input to obtain The deep neural network model; secondly, establish the migration fusion network, and migrate the network parameters of the deep neural network model to the migration fusion network; then use the one-dimensional sequence signal samples to train the migration fusion network to obtain the migration fusion model; finally, use the migration The fusion model recognizes the one-dimensional sequence signal collected in real time, and outputs the fault category of the one-dimensional sequence signal. The invention proposes a two-level migration mechanism, through which multi-source heterogeneous information is fused, which can not only fuse the features of one-dimensional sequence signals and screenshot images to improve the accuracy of fault diagnosis, but also avoid convolution operations and reduce the Time complexity improves the real-time performance of fault diagnosis and achieves the purpose of real-time fault diagnosis.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis, in particular to a multi-source heterogeneous information fusion fault diagnosis method based on two-level migration learning. Background technique [0002] In the actual equipment monitoring, key sensors are monitored and the monitoring center monitor is used to present the dynamic change curve of the values ​​collected by some key sensors. These monitoring screenshots can show the amplitude and change trend of the vibration signal over a period of time, so the obtained data There are both one-dimensional signal sequences and two-dimensional screen shots. If only one type of data in multi-source heterogeneous data is used for deep learning fault diagnosis, the information contained in other data will be wasted, which will affect the accuracy of fault diagnosis. Existing fusion methods use deep neural networks and convolutional neural networks to extract multi-source heterogeneous featur...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problem of poor real-time performance of the fault diagnosis algorithm when multi-source heterogeneous information and external domain data are fully utilized, the present invention proposes a multi-source heterogeneous information fusion fault diagnosis method based on two-level transfer learning. The fusion of multi-source heterogeneous information can not only fuse the characteristics of two-dimensional images to improve the accuracy of fault diagnosis, but also avoid convolution operation, improve the real-time performance of fault diagnosis, and achieve the purpose of real-time fault diagnosis

Images