To assess health conditions of rotating machinery efficiently, multiple accelerometers are mounted on different locations to acquire a variety possible faults signals. The statistical features extracted from these signals identify the running status machine. However, acquired vibration due sensor's arrangement and environmental interference, which may lead diagnostic results. In order improve fault diagnosis reliability, new multisensor data fusion technique is proposed. First, time-domain...

Deep neural networks present very competitive results in mechanical fault diagnosis. However, training deep models require high computing power while the performance of architectures extracting discriminative features for decision making often suffers from lack sufficient data. In this paper, a transferable convolutional network (CNN) is proposed to improve learning target tasks. First, one-dimensional CNN constructed and pretrained based on large source task datasets. Then transfer strategy...

Recently, deep learning-based intelligent fault diagnosis techniques have obtained good classification performance with amount of supervised training data. However, domain shift problem between the and testing data usually occurs due to variation in operating conditions interferences environment noise. Transfer learning provides a promising tool for handling cross-domain problems by leveraging knowledge from source help target domain. Most existing studies attempt learn both features common...

Intelligent compound fault diagnosis of rotating machinery plays a crucial role for the security, high-efficiency, and reliability modern manufacture machines, but identifying decoupling are still great challenge. The traditional methods focus on either bearing or gear diagnosis, where is always regarded as an independent pattern in process relationship between single not considered completely. To solve such problem, novel method called deep convolutional neural network proposed intelligent...

Recently, domain adaptation techniques have achieved great attention in solving domain-shift problems of mechanical fault diagnosis. However, existing methods mostly work under assumption that source and target share identical label spaces, which fail to handle those issues, where a large set data classes are available only cover subset classes. To address this problem, novel weighted adversarial transfer network (WATN) is proposed for partial diagnosis, article. Adversarial training...

Recently, deep transfer learning based intelligent fault diagnosis has been widely investigated, and the tasks that source target domains share same categories have well addressed. However, due to complexity uncertainty of mechanical equipment, unknown new faults may occur unexpectedly. This problem received less attention in current research, which seriously limited application learning. In this article, a two-stage adversarial network is proposed for multiple detection rotating machinery....

In recent years, deep learning has become a promising tool for rotary machinery fault diagnosis, but it works well only when testing samples and training are independent identically distributed. practice, usually under variable speed. The change of speed leads to the variation samples' distribution, which can significantly decrease performance model. Scholars try utilize transfer techniques solving this problem. However, most exiting methods just work target instead all speed, while always...

In real industries, there often exist application scenarios where the target domain holds fault categories never observed in source domain, which is an open-set adaptation (DA) diagnosis issue. Existing DA methods under assumption of sharing identical label space across domains fail to work. What more, labeled samples can be collected from different sources, multisource information fusion rarely considered. To handle this issue, a approach developed. Specifically, data operation conditions...

Generally, high performance of deep learning (DL)-based machinery fault diagnosis methods relies on abundant labeled samples under various working conditions, while they are usually stored by different users and not communicated with each other due to data privacy protection. Federated (FL) is a possible solution, but the traditional federated averaging (FedAvg) algorithm in FL ignores potential domain shift participants, which limits its further application. Therefore, transfer framework...

With significant advantages in feature learning, the deep learning based compound fault diagnosis method has brought many successful applications for industrial equipment. However, few studies focus on interpretability of intelligent methods, and results are hard to interpret which prevents wide application these methods practical scenarios. To solve above challenging problems, an interpretable framework, called wavelet capsule network (WavCapsNet), is proposed machinery by leveraging...

Compound fault, as a primary failure leading to unexpected downtime of rotating machinery, dramatically increases the difficulty in fault diagnosis. To deal with encountered implementing compound diagnosis (CFD), researchers and engineers from industry academia have made numerous significant breakthroughs recent years. Admittedly, many systematic surveys focused on been conducted by reputable researchers. Nevertheless, previous review articles paid more attention several single or...

With advanced measurement technologies and signal analytics algorithms developed rapidly, the past decades have witnessed large amount of successful breakthroughs applications in field intelligent fault diagnosis (IFD). However, historical IFD methods difficulties for compound diagnosis, when labeled data cannot be collected advance new or extreme working conditions. Facing with such challenges, a deep adversarial capsule network (DACN) is proposed to embed multidomain generalization into so...

Deep transfer learning has attracted many attentions in machine intelligent fault diagnosis. However, most existed deep algorithms encounter difficulties to detect a new emerging target domain because these methods assume that the source and domains have same categories. Unfortunately, real-world applications, may emerge during running, which is not as those faults for training diagnosis models. To solve this problem, novel method named adversarial network (DATLN) proposed detection. First,...