An autoencoder with adaptive transfer learning for intelligent fault diagnosis of rotating machinery

Abstract

Under variable working conditions, a problem arises, which is that it is difficult to obtain enough labeled data; to address this problem, an adaptive transfer autoencoder (ATAE) is established to diagnose faults in rotating machinery. First, a data adaptation module, which calculates the maximum mean discrepancy for the network hidden-layer data in reproducing kernel Hilbert space, is introduced to the autoencoder network, thus making the classification model operate under variable working conditions. Variation particle-swarm optimization is then invoked to optimize the data adaptation parameters. Finally, the k-nearest neighbors algorithm, as the classification layer of the network, identifies the state of health of the rotating machinery. The capabilities of the intelligent fault-diagnosis network are verified using vibration signals from a bearing test rig and a gearbox test rig. The experimental results suggest that, compared with state-of-the-art diagnosis methods, the proposed ATAE network can significantly boost diagnostic performance in the absence of target vibration signal labels.

Local parameter identification with neural ordinary differential equations

“…However, data-driven approaches can obtain hidden knowledge from experimental data and infer the current system state for PHM [19,[23][24] . Convolutional neural network (CNN) and recurrent neural network (RNN) have led to many meaningful studies [19,[25][26][27][28][29] , such as the establishment of an adaptive transmission autoencoder (ATAE) for rotating machinery fault diagnosis [25] and the proposal of a deep morphological convolutional network (DMCNet) for feature learning of gearbox vibration signals [26] . These are all excellent applications of the data-driven approach.…”

Weighted asynchronous subdomain adaptation network for unsupervised fault diagnosis of bearing

“…Bearing plays an important role in engineering fields, and its health states have a crucial impact on the safety of the equipment [1]. Early fault diagnosis of bearing is mainly based on signal processing methods and is still improving continuously [2][3][4].…”

Gearbox Fault Diagnosis Based on Two-Class NMF Network Under Variable Working Conditions

“…In [19], an unsupervised multi-view sparse filtering method was proposed to filter the interfering electrical signals while improving the accuracy of fault diagnosis. In view of the limited data of gearbox faults and the lack of an accurate model of the actual gearbox, in [20,21], an automatic encoder model is used to process the limited sample data to achieve accurate diagnosis of gearbox faults. In [22], healthy data is used to supplement the consistency of the ordered frequency spectrum, and then the probabilistic model is used for automatic fault detection and calculation of diagnostic indicators for positioning, which reduces the requirements for fault data.…”