No abstract

Bearing failure is the most common failure mode in rotating machinery and can result in large financial losses or even casualties. However, complex structures around bearing and actual variable working conditions can lead to large distribution difference of vibration signal between a training set and a test set, which causes the accuracy-dropping problem of fault diagnosis. Thus, how to improve efficiently the performance of bearing fault diagnosis under different working conditions is always a primary challenge. In this paper, a novel bearing fault diagnosis under different working conditions method is proposed based on domain adaptation using transferable features(DATF). The datasets of normal bearing and faulty bearings are obtained through the fast Fourier transformation (FFT) of raw vibration signals under different motor speeds and load conditions. Then we reduce marginal and conditional distributions simultaneously across domains based on maximum mean discrepancy (MMD) in feature space by refining pseudo test labels, which can be obtained by the nearest-neighbor (NN) classifier built on training data, and then a robust transferable feature representation for training and test domains is achieved after several iterations. With the help of the NN classifier trained on transferable features, bearing fault categories are identified accurately in final. Extensive experiment results show that the proposed method under different working conditions can identify the bearing faults accurately and outperforms obviously competitive approaches.

Most fault detection methods based on the assumption of working in stationary or approximate stationary conditions are limited under varying operation conditions, for that the frequency aliasing phenomenon is inevitable in the spectrum. Therefore, in order to handle the problem of fault diagnosis under non-stationary conditions, researchers have proposed numerous methods and some achievements have been obtained. In this article, a new feature extraction method is proposed for fault diagnosis of rolling bearings under varying speed conditions. Based on the assumption that the energy will increase when balls cross over fault position, frequency values are divided by instantaneous speed and arranged in the descending order of corresponding amplitude to form a new fault feature array, that is, the ratio of frequency to instantaneous speed reconfiguration arrays. Thereafter, the Euclidean distance classifier is utilized for recognition. The efficacy of the proposed method is demonstrated by simulated and experimental data. Categorized results show that the new approach is capable of handling the bearing fault classification under varying speed conditions.

Bearing faults are the most common failure modes in the rotating system. Vibration data from the rotating system carry important information, that is, characterization and diagnosis; therefore, the vast vibration signals collected from multiple sensors mounted in different sites are transmitted in a certain order for online fault diagnosis. However, due to the influence of transfer paths and noises, the sensitivities to the same fault signal of measured data streams are of significant differences, and signals containing weak sensitivity to the fault are likely to be transmitted preferentially while neglecting transmission order. Meanwhile, high volume vibration data greatly increase the transmission burden. These above-mentioned reasons dramatically reduce online diagnostic efficiency. Thus, fully considering the sensitive differences to the fault for multiple channels, how to transmit measured data streams of multiple sensors for timely online detecting the bearing failure is still a primary challenge. In order to solve this problem, a novel online bearing fault diagnosis method based on the multiple data streams transmission schemes (MDSTS) is proposed in this paper. Multiple sensors are numbered consecutively, and data streams from all channels are transmitted according to the preset order and transport protocol via a certain length at the beginning of diagnosis. Then, a fault sensitivity assessment model (FSAM) is established on maximum mean discrepancy (MMD) for transmitting the most sensitive data stream by calculating the distribution discrepancies between each channel's data streams and the historical datasets in the frequency domain, and then, the fault diagnosis model based on K-nearest neighbor (KNN) trained on historical datasets was used to evaluate the transmission scheme and acquire reliable diagnostic results via predicting performances of multiple and consecutive datablocks until all these exceed an alarm value. The extensive experiment results show that the proposed method can timely and accurately identify the bearing faults and outperforms obviously competitive approaches.

Traditional data-driven intelligent fault diagnosis methods have been successfully developed under the closed set assumption (CSA). CSA-based fault diagnosis assumes that the fault types in the test set are consistent with that in the training set, which can achieve high accuracy, but this is generally not valid in real-world industrial applications where the collection of data in industrial applications is often limited. As it is unrealistic to assume that the training set will cover all fault types, the application of the fault classifier may fail when the test set contains unknown fault types because the probability of input samples belonging to unknown types cannot be obtained. To solve the problem of how unknown fault types may be accurately identified, this paper further studies the open set assumption (OSA) fault diagnosis. We propose an open set convolutional neural network (OS-CNN) method and apply our OS-CNN model to an improved OpenMax method as a deep network to accurately detect unknown fault types. The overall performance was significantly improved as our OS-CNN model was able to effectively tighten the boundary of known classes and limit the open-space risk for the OpenMax method based on distance modeling. The overall effectiveness of the proposed method was verified by experimental studies based on four different bearing datasets. Compared with state-of-the-art OSA fault diagnosis method, our method cannot only realize the correct classification of the known fault classes, but it can also accurately detect the unknown fault classes.