Remaining useful life (RUL) prediction of rolling element bearings plays a pivotal role in reducing costly unplanned maintenance and increasing the reliability, availability, safety machines. This paper proposes hybrid prognostics approach for RUL bearings. First, degradation data are sparsely represented using relevance vector machine regressions with different kernel parameters. Then, exponential models coupled Fréchet distance employed to estimate adaptively. The proposed is evaluated...

The success of intelligent fault diagnosis machines relies on the following two conditions: 1) labeled data with information are available; and 2) training testing drawn from same probability distribution. However, for some machines, it is difficult to obtain massive data. Moreover, even though can be obtained method trained such possibly fails in classifying unlabeled acquired other due distribution discrepancy. These problems limit successful applications As a potential tool, transfer...

The remaining useful life (RUL) prediction of rolling element bearings has attracted substantial attention recently due to its importance for the bearing health management. exponential model is one most widely used methods RUL bearings. However, two shortcomings exist in model: 1) first predicting time (FPT) selected subjectively; and 2) random errors stochastic process decrease accuracy. To deal with these shortcomings, an improved proposed this paper. In model, adaptive FPT selection...

Remaining useful life (RUL) prediction allows for predictive maintenance of machinery, thus reducing costly unscheduled maintenance. Therefore, RUL machinery appears to be a hot issue attracting more and attention as well being great challenge. This paper proposes model-based method predicting machinery. The includes two modules, i.e., indicator construction prediction. In the first module, new health named weighted minimum quantization error is constructed, which fuses mutual information...

Remaining useful life (RUL) prediction is a key process in condition-based maintenance for machines. It contributes to reducing risks and costs increasing the maintainability, availability, reliability, productivity of This paper proposes new method based on stochastic models machine RUL prediction. First, model constructed considering multiple variability sources degradation processes simultaneously. Then Kalman particle filtering algorithm used estimate system states predict RUL. The...

Deep transfer-learning-based diagnosis models are promising to apply knowledge across related machines, but from which the collected data follow different distribution. To reduce distribution discrepancy, Gaussian kernel induced maximum mean discrepancy (GK-MMD) is a widely used distance metric impose constraints on training of models. However, using GK-MMD have three weaknesses: 1) may not accurately estimate because it ignores high-order moment distances data; 2) time complexity high...

To integrate the complete degradation information of machinery, deep learning-based prognostics approaches usually use monitoring data acquired by different sensors as inputs networks. These approaches, however, lack an explicit learning mechanism to effectively identify distinctions sensor and highlight important information, thereby affecting accuracy networks limiting their generalization. overcome aforementioned weaknesses, a new framework named multiscale convolutional attention network...

Due to the data distribution discrepancy, fault diagnosis models, trained with labeled in one scene, likely fails classifying by unlabeled acquired from other scenes. Transfer learning is capable generalize successful application scene However, existing transfer methods do not pay much attention reduce adaptively marginal and conditional biases, also ignore degree of contribution between both biases among network layers, which limit classification performance generalization reality. To...

Remaining useful life (RUL) prediction has attracted more and attention in recent years because of its significance predictive maintenance. The degradation processes systems from the same population are generally different one another due to their various operational conditions health states. This behavior is defined as unit-to-unit variability (UtUV), which brings difficulty RUL prediction. To handle this problem, paper develops a Wiener-process-model (WPM)-based method for with...

摘要: 预测与健康管理对保障机械装备安全服役、提高生产效率、增加经济效益至关重要。高质量的全寿命周期数据是预测与健康管理领域的基础性资源,这些数据承载着反映装备服役性能完整退化过程与规律的关键信息。然而,由于数据获取成本高、存储与传输技术有待发展等原因,典型的全寿命周期数据极其匮乏,严重制约了机械装备预测与健康管理技术的理论研究与工程应用。为解决上述难题,西安交通大学机械工程学院雷亚国教授团队联合浙江长兴昇阳科技有限公司,选取工业场景中典型的关键部件——滚动轴承为试验对象,开展了历时两年的滚动轴承加速寿命试验,并将获取的试验数据——XJTU-SY滚动轴承加速寿命试验数据集面向全球学者公开发布。该数据集共包含3种工况下15个滚动轴承的全寿命周期振动信号,采样频率高、数据量大、失效类型丰富、记录信息详细,既可为预测与健康管理领域提供新鲜的"数据血液",推动故障诊断与剩余寿命预测等领域的算法研究,又可助力工业界智能化运维的"落地生根"。

Variational mode decomposition provides a feasible method for non-stationary signal analysis, but the is still not adaptive, which greatly limits wide application of method. Therefore, an adaptive spectrum extraction proposed in this article. The mainly composed spectral segmentation, extraction, and feedback adjustment. In segmentation part, considering lack robustness classical scale space strong noise environment, article proposes fault feature mapping, solves over-decomposition...

In recent years, intelligent data-driven prognostic methods have been successfully developed, and good machinery health assessment performance has achieved through explorations of data from multiple sensors. However, existing data-fusion approaches generally rely on the availability all sensors, are vulnerable to potential sensor malfunctions, which likely occur in real industries especially for machines harsh operating environments. this paper, a deep learning-based remaining useful life...

Event-based cameras are the emerging bioinspired technology in vision sensing. Different from traditional standard cameras, event-based asynchronously record brightness change per pixel, and have great merits of high temporal resolution, low energy consumption, dynamic range, etc. While been initially exploited several common vision-based tasks recent years, investigation on machine condition monitoring problem is quite limited. This article offers first attempt current literature exploring...

As the fundamental and key technique to ensure safe reliable operation of vital systems, prognostics with an emphasis on remaining useful life (RUL) prediction has attracted great attention in last decades. In this paper, we briefly discuss general idea advances various RUL methods for machinery, mainly including data-driven methods, physics-based hybrid etc. Based observations from state art, provide comprehensive discussions possible opportunities challenges machinery so as steer future...

Dear Editor, This letter presents a novel dynamic vision enabled contactless cross-domain fault diagnosis method with neuromorphic computing. The event-based camera is adopted to capture the machine vibration states in perspective of vision. A specially designed bio-inspired deep transfer spiking neural network (SNN) model proposed for processing event streams visionary data, feature extraction and diagnosis. can also extract domain-invariant features from different operating conditions...