Anomaly detection is one of the most fundamental and indispensable components in predictive maintenance. In this article, anomaly modeled as a one-class classification problem. Based on scenario that training data only include healthy state data, fault-attention generative probabilistic adversarial autoencoder (FGPAA) proposed to automatically find low-dimensional manifold embedded high-dimensional space signal. Benefited from characteristics autoencoder, signal information loss feature...

Recent researches on intelligent fault diagnosis algorithms can achieve great progress. However, considering the practical scenarios, amount of labeled data is insufficient in face difficulty annotation, which would raise risk overfitting and hinder model from its industrial applications. To address this problem, article, we propose an interinstance intratemporal self-supervised learning framework, where massive unlabeled integrated with supervised few to enrich capacity learnable data....

Remaining useful life (RUL) prediction plays a vital role in prognostics and health management (PHM) for improving the reliability reducing cycle cost of numerous mechanical systems. Deep learning (DL) models, especially deep convolutional neural networks (DCNNs), are becoming increasingly popular RUL prediction, whereby state-of-the-art results have been achieved recent studies. Most DL models only provide point estimation target RUL, but it is highly desirable to associated confidence...

Surface damage detection is vital for diagnosis and monitoring of aeroengine blade. At present, borescope inspection the dominant technology. Several inspectors hold to inspect blades one by through naked eyes on apron. The turbine even requires drilling into narrow tail nozzle. manual visual high cost low efficiency. To improve efficiency economic benefit, we propose an intelligent method in this article. Facing problem weak information caused background noise unsatisfactory illumination,...

As the core part of Prognostic and Health Management (PHM) major equipment such as high-speed trains aero engines, bearing fault classification have been research priorities in field. Although convolutional neural network (CNN) has shown good results this type task, real application with limited training data makes CNN a big gap between actual expected effect. Therefore, faults class-imbalance is very practical work. In paper, semi-supervised information maximizing generative adversarial...

Intelligent anomaly detection methods have achieved much successes in machinery condition monitoring. However, the underlying independent and identically distributed assumption restricts their application scopes to steady operating conditions. False missing alarms would occur when machines operate under time-varying circumstances. In this work, a more challenging setting is studied, where working conditions are continuously changing, such that few or no samples available for model training...

Intelligent anomaly detection (AD) methods have achieved much successes in machinery condition monitoring. However, the underlying independent and identically distributed assumption restricts their application scopes to steady operating conditions. False missing alarms would occur when machines operate under time-varying circumstances. In this work, a more challenging setting is studied, where working conditions are continuously changing, such that few or no samples available for model...

Precise location of laser spot in precision measurement is always an important research direction. Laser has the characteristics good direction and small divergence, so it widely used aerospace, weapon systems optical measuring testing instruments. The accuracy center can directly determine measurement. Aiming at positioning spot, foundation researching limitation practical application common algorithm, this paper proposes a method localization based on sub-pixel interpolation, which...

Abstract With the rapid development of display technology, human beings have more and stringent requirements for products, e.g., their light weight, thinness, flexible performances, versatility. Contemporarily, organic light-emitting diode (OLED) technology its derived multi-layer structure tend to be wearable, light, responsive in a short time. Although OLED is popular, there are still many problems that needs addressed, new raw materials, rough substrate, low efficiency workmanship,...

This paper rigorously establishes that, by increasing the number of dendritic branches, we can achieve an infinite approximation to any continuous function with accuracy. In other words, accuracy be improved using one neuron more branches under premise backpropagation neural network theory.

Fault diagnosis is of vital importance to maintain safety and reliability mechanical equipment. Intelligent diagnostic methods have achieved high performance in recent researches. However, industrial application, machines will suffer complex noises the operating condition varying as well, which leads domain shift degradation. As a promising alternative supervised learning, self-supervised contrastive learning follows paradigm extract robust feature representation. Nevertheless, training...

Non-line-of-sight (NLOS) imaging technology is to 'see' the target out of sight, such as an object around a corner or hidden by some shelters. However, due constraints device definition and computing load, NLOS system usually expensive requires objects with special material simple shape. Besides, space limited. We perform series simulation 1550nm infrared laser expand application field improve performance system. Based on math physical properties, main experimental components are modeled...