This paper presents a new unsupervised anomaly detection approach for spacecraft based on normal behavior clustering. method takes as input set of unlabelled historical telemetry data and automatically detects anomalies within the data. After these abnormal are removed, constructs system model Then at run-time, it monitors status any appearing in real-time by checking deviations from model. The experimental results show that is efficient practical system.

Reducing the dimension of hyperspectral image data can directly reduce redundancy data, thus improving accuracy classification. In this paper, deep belief network algorithm in theory learning is introduced to extract in-depth features imaging spectral data. Firstly, original mapped feature space by unsupervised methods through Restricted Boltzmann Machine (RBM). Then, a will be formed superimposed multiple Machines and training model parameters using greedy layer layer. At same time, as...

Abstract We are concerned with the inverse scattering problems associated incomplete measurement data. It is a challenging topic of increasing importance that arise in many practical applications. Based on prototypical working model, we propose machine learning based scheme, which integrates CNN (convolution neural network) for data retrieval. The proposed method can effectively cope reconstruction under limited-aperture and/or phaseless far-field Numerical experiments verify promising...

Abstract Ultra-high voltage AC substation has the characteristics of high level, large area, complex equipment structure, heavy inspection task and operation risk. The existing video surveillance system is mainly used for security, but distribution points insufficient clarity low, so it can’t achieve full coverage points. UHV needs to develop remote intelligent assist maintenance staffs carry out daily work, reduce workload staffs, improve speed emergency response ensure safety operation. In...

8 inch 4H-silicon carbide (SiC) development faces challenges first from obtaining high-quality SiC seed substrate, then reducing grown-in crystal residual stress and defects in the following growth process. Here we report diameter expansion process 6 4H-SiC substrate to crystal. Based on simulation experimental results, it is deduced that an optimized radial temperature gradient (RTG) zone range of 0.10-0.12 °C/mm essential for efficient expansion. According RTG calculation, designed as well...

Abstract Distributed smart grid is an organic carrier for achieving renewable energy substitution and promoting new consumption. It important component of the construction power systems. To meet current operational needs distributed construction, a protection method based on AC/DC hybrid distribution network proposed. The application proposed in this article, has good promotion value.

The harmonic problem caused by HVDC transmission has always been the focus of research on amplification at receiving-end power grid. increase in proportion cable lines grid makes more complicated. This paper puts forward concept rate, and derives model line that can accurately account for rate. better consider two outgoing modes DC inverter substations. Finally, voltage transfer coefficient from system to is established, detailed calculations are carried out based this model. results show...

In order to meet the reliability requirements of Xinjiang EHV power grid, improve switching operation efficiency substation, and reduce risk misoperation, this paper determines sequence control judgment based on different modes primary equipment, proposes design application one key in substation by comparing technology economy sensors, combining with anti misoperation host intelligent technology. On premise ensuring safety, it can effectively outage time, rate, thus accident rate prevent...

Time series data is very popular and common in the world, which widely produced various real time monitoring systems, such as satellites, power plants, cardiac electrophysiology, financial transactions. In these applications, it a critical problem to precisely recognize pattern of random window on growing stream. Since has stochastic phase shift with standard known pattern. That leads great recognition error for existing methods. This paper presents Multi-channel-scale Convolutional Neural...

Real-time rendering of large-scale and complex scenes is one the important subjects in virtual reality technology. In this paper, we present an efficient view-dependent out-of-core algorithm for scenes. preprocessing phase, partition scene hierarchy compute continuous hierarchical level detail (HLOD) each node using a dynamic topology simplification method. Then at run-time, multi-threaded technique used. The thread uses coarse global refinement HLODs fine local refinement. prefetching...

With the development and application of computer science technology innovation, artificial intelligence industry has achieved sharp development, its been widely used in many fields, such as natural language processing, robot manufacturing, semantic recognition, information system etc. In terms recognition understanding human based on become focus research. And processing play an important part they have satisfactory results Abstract Meaning Representation (AMR) Vector Space Model (VSM)...

We are concerned with the inverse scattering problems associated incomplete measurement data. It is a challenging topic of increasing importance in many practical applications. Based on prototypical working model, we propose machine learning based scheme, which integrates CNN (convolution neural network) for data retrieval. The proposed method can effectively cope reconstruction under limited-aperture and/or phaseless far-field Numerical experiments verify promising features our new scheme.