Search

The number of labeled samples has a great impact on the classification results of a very-high-resolution (VHR) remote sensing image. However, the acquisition of available labeled samples is difficult and time consuming. Faced with the limited labeled samples on a high-resolution remote sensing image, a semisupervised method becomes an effective way. In semisupervised learning, an accurate similarity prediction between unlabeled and labeled samples is very important. However, a reliable similarity prediction between high-dimensional features is difficult. For more reliable similarity prediction for the high-dimensional feature, a novel semisupervised classification framework via improved metric learning with a convolutional neural network is proposed. In the proposed method, a novel trainable metric learning network is designed to accurately evaluate the similarity between high-dimensional features. The vector distance parameter solving problem is transformed into a neural network design problem, which can automatically calculate parameters by the back-propagation algorithm. Finally, the pixel constraint mechanism is introduced to select the unlabeled samples. Experimental results conducted on three VHR remote sensing images, including Aerial, Xi’an, and Pavia University, and the results present that the proposed method performs better than the compared state-of-the-art methods.

Pitting corrosion of rotating machinery is one of the most common faults in industrial engineering. The convolutional neural network (CNN) is increasingly applied to the fault diagnosis. However, the conventional CNN method will reduce the feature dimension of the collected signal and cause the loss of information during the pooling process. In this paper, a new method based on dilated CNN combined with spatial dropout (DCSD) is proposed to diagnose the early faults of rotating machinery. By filling the convolution kernel, the DCSD method can increase the receptive field of the CNN without increasing the number of parameters while retaining more features of the raw vibration signal of the rotating machine. To avoid the dropout method eliminates the adjacent elements with a strong correlation, the Spatial Dropout method is adopted to reduce the overfitting problem of deep networks. The early pitting gears experiment was designed to verify the DCSD method in this paper. The raw vibration signal data of 6 different healthy states were collected to verify the effectiveness of the method. The experimental results show that the DCSD method proposed can effectively distinguish the different early gears pitting, and the diagnostic accuracy is better than other popular deep learning methods.

Land cover change detection (LCCD) based on bitemporal remote sensing images has become a popular topic in the field of remote sensing. Despite numerous methods promoted in recent decades, an improvement on the usability and performance of these methods has remained necessary. In this paper, a novel LCCD approach based on the integration of $k$ -means clustering and adaptive majority voting ( $k$ -means_AMV) techniques have been developed. The proposed $k$ -means_AMV method consists of three major techniques. First, to utilize the contextual information in an adaptive manner, an adaptive region around a central pixel is constructed by detecting the spectral similarity between the central pixel and its eight neighboring pixels. Second, when the extension for the adaptive region is terminated, the $k$ -means clustering method is applied to determine the label of each pixel within the adaptive region. Finally, an existing AMV technique is used to refine the label of the central pixel of the adaptive region. When change magnitude image (CMI) is scanned and processed in this manner, the label of each pixel in the CMI can be refined and the binary change detection map can be generated. Three image scenes related to different land cover change events are adapted to test the effectiveness and performance of the proposed $k$ -means_AMV approach. The results show that the proposed $k$ -means_AMV approach demonstrates better detection accuracies and visual performance than that of the several extensively used methods.

In recent years, machine learning is applied in the bioinformatics and medical fields to analyze relationships among biological features and behaviors. However, it is difficult to discover the significant features of large-scale datasets. A novel feature extraction method called bioinformatic explainable deep neural network (BioExpDNN) is proposed to filter the critical features with strong influences on the dataset and to explain the interaction of features. In the practical experiments, this study adopted three biomedical science datasets from the UCI (University of California, Irvine) Machine Learning Repository: (1). Cryotherapy Data Set (CDS) contains 6 attributes and 2 classes (i.e., recovery and non-recovery); (2). Cervical Cancer Behavior Risk Data Set (CCBRDS) consists of 18 attributes and 2 classes (i.e., cervical cancer patient and healthy body); (3). Heart Failure Clinical Records Data Set (HFCRDS) includes 12 clinical attributes and 2 classes (i.e., death and life). In comparison results, extracted features were considered as inputs of a classifier based on deep neural network for classification. The classification accuracy was selected as an evaluation factor to evaluate the performance of feature extraction methods. The experimental results showed that the classification accuracies of CDS, CCBRDS, and HFCRDS were 92.59%, 100%, and 78.9%, respectively.

Stretchable solar arrays meet the demands of light weight and flexibility for future solar array design. Research in stretchable electronics has uncovered ways to build stretchable systems with inorganic materials. Island-bridge layout and structured substrate are two effective methods to build stretchable systems. This work proposes a zigzag structured substrate for stretchable solar arrays to provide levels of stretchability.A zigzag structure serves as the device island of the solar array to absorb and isolate large deformation of the substrate. An analytic model is established to analyze the stretchability of a planar zigzag structure. A stretchable solar array with a zigzag structured substrate is analyzed by FEM.The solar array with a zigzag structured substrate enables high area coverage and stretchability. This work presents a new approach for design of stretchable solar array and introduce a simply effective model as guidelines.

In oder to solve the problems of orderly control of distributed renewable energy and electric vehicles(EV) in the distribution network, this paper studies the orderly energy management of distributed renewable energy and EV. The probability distribution of wind power and solar power are analyzed, and the equivalent load model of wind-solar power is established by probabilistic method. Meanwhile, the analysis based on random travel simulation is carried out, and model of EV is established as well, which refinines and completing the structure methods of the travel feature. At the same time, the three-layer energy management model of the EV is constructed, the duties and task division between network dispatching center and charging service provider are clarified, as well proposing the energy management strategy of tight coupling and loosely coupling. The simulation results show the energy management model proposed in this paper can effectively promote renewable energy consumption and load smoothing, which meets the needs of charging service providers through receiving guidance and self-regulation of EV in their jurisdiction area.

Convolutional neural networks (CNNs) have proven to be an effective way for deep feature extraction. However, multispectral and panchromatic images are susceptible to illumination unevenness and noise, and the default cross entropy loss function consider only the local information, resulting in misclassification. In this paper, we propose a novel super-pixel-level deep neural networks for multispectral and panchromatic images classification, and define a novel percep-tualloss function via non-local spectral and structure similarity to suppress the interference of unbalanced light and noise. We also propose the corresponding iteration optimization algorithm in this paper. Experimental results show that the proposed method performs better than the state-of-the-art methods.

The risk assessment of power network is beneficial to improve the reliability of power system. It only considers the condition of the equipment itself in conventional power network risk assessment methods, but lacks of a consideration of the various influence factors in the network. A comprehensive risk assessment method based on Equipment Condition and entropy is proposed to evaluate power network in this paper. Firstly, a model of power network is established, considering the key risk factors including equipment condition, network structure, loads, natural and weather etc. Then, the AHP method is used to determine the subjective weight, while entropy method is applied to obtain the objective weights. After that, the combined weights are calculated based on subjective and objective weights. Last, the effectiveness of the model is proved by two study cases, the weights parameters in the model can be applied in the other risk assessment of power networks.

As a traditional approach of partial discharge (PD) in power transformers, ultrasonic detection with piezoelectric transducers (PZT) has drawbacks of low sensitivity and low detection efficiency. In order to overcome these disadvantages, an ultrasonic detection system based on fiber Bragg grating (FBG) is developed for PD in power transformers. Take the advantage of good insulation ability and immunity to electromagnetic interference, the optical sensor can be put into the transformers and detect the weak PD signal directly. In this paper, ultrasonic detection theory of FBG was analyzed, and PD detection experiments in transformer oil were carried out to test the performance of this FBG ultrasonic detection system. Furthermore, in order to demonstrate the coherence and distributed measuring ability, PD detection experiments based on multi-point FBG string with different center wavelength were conducted. Study results show the performance of FBG ultrasonic detection system is better than traditional PZT sensor, and satisfactory detection coherence of FBG with different center wavelength is proved. Moreover, distributed ultrasonic measuring of PD in power transformers can be realized by multi-point FBG string with application of wavelength-division and time-division multiplexing technology.

Multi-objective evolutionary algorithms (MOEAs) can be mainly divided into set approximation methods and decomposition methods. The former approximates the Pareto front by the whole population directly, while the latter solves decomposed subproblems. The theoretical understanding of these methods is, however, quite insufficient. In this paper, we try to gain more understanding by investigating a combination of set approximation MOEAs with a sequential decomposition mechanism. Our theoretical analysis shows that, the combination achieves a better running time than the corresponding set approximation MOEAs by a factor n (the problem size) on synthetic problems as well as the minimum spanning tree problem, which hints that the two types of MOEAs might be mutually complemental.

Precise timing by the technique of GNSS common view is proven to be an effective technique with the accuracy of nanosecond level in real-time mode. The timing technique of carrier phase is also applied in post mode with the accuracy of sub-nanosecond level, but the real-time mode still has little explicit research. In this study, we develop the precise timing algorithm of carrier phase to investigate timing performance o in real-time mode. The four stations from the International GNSS Service (IGS) data network are processed, which are equipped high performance external time frequency. And then three kinds of satellite orbit and clock products with its corresponding accuracy and latency are applied into the timing algorithm for investigate the timing accuracy with different time latency. The RMS is achieved with 0.35ns level by the IGS rapid satellite orbit and clock product, and then the prediction of the IGS ultra rapid product is about 0.52ns.

In the mobile social networks, users are used to depending on mobile network in daily life. Users can produce lots of records. Points of interests (POIs) are the description of the users' daily behavior. In this paper, we focus on protection against disclosure in the publication of POIs with time information. There are some protection techniques. In general, to protect privacy of data, the common means are generalization and suppression. In fact, adding appropriate noise nodes is also a good idea. Some other papers distinguish the sensitive nodes and the non-sensitive nodes. In our paper, we take time information into consideration because POIs and their corresponding time information are practical utility in applications. In addition, for different conditions and different individuals, the sensitive part of POIs is different. In order to make the published data more practical, we need the data to be as complete as possible in POIs and time information. We design a method of partitioning which guarantees km-anonymity for the privacy protection.

Inverse Synthetic Aperture Radar has been proved to be a powerful tool in imaging moving targets in military and civilian applications. The target's movement with respect to the radar sensor provides the angular diversity required for range-Doppler ISAR imagery and image blur due to its motion error. For real targets such as plane, ship and tank have complicated motion components which are unknown to the radar. This paper is a brief review of universally used motion compensation techniques employed in ISAR for a clear and focused image.

In this paper, a method for removing ultrasound medical image noise is proposed. This method improves the original median filter by designing a suit of directional templates. In order to obtain local directions, a filter, which combines Gaussion blur with direction parameter, is designed. Then, directional templates based on local directions mentioned above are used as the filter for reducing spark noise. At last, the finial denoising result of each pixel is obtained from the filtered image or the original noise-polluted image by judging their differences. If the difference between two corresponding pixels of the two images is less than a threshold, the current pixel value of the noise-polluted image is kept. Otherwise, the one of the filtered image is used. The proposed method preserves the contour and the texture areas, which increases visual effects. Experiment results show the proposed method achieves sound performance on the synthetic images created by Field II.

Ultrasonic examination is a routine inspection technology. It has several merits, such as no harm to human body, cheap and relative high precision inspection. So it is widely used in physical examination and various types of organ inspections. In order to increase the detection rate of liver disease in ultrasound images, a method extracting the liver region from ultrasound images is proposed in this paper. This method firstly deals with uneven illumination of ultrasound image, which makes the brightness of liver region in images to be consistent. Then, in order to better resist the noise, the Fuzzy C Mean (FCM) method using the priori shape information, which is called FCM_I, is proposed to segment the image. Finally, according to the distribution and shape of the liver, the largest foreground area in the image is obtained. The proposed method obtains good results in the abdominal ultrasound images obtained by the hospital.

Multiscale fuzzy entropy (MFE) is an effective algorithm which has been successfully applied in many fields for measuring the complexity of a time series. Though, MFE can yield inaccurate entropy estimations as the coarse-graining procedure used by the algorithm reduces the length of the time series under investigation. A modified multiscale fuzzy entropy (MMFE) algorithm is presented in this paper to overcome this problem. In this new approach, the coarse-graining procedure is replaced by a moving-average procedure which constructs template vectors in calculating the fuzzy entropy. The effectiveness of the proposed MMFE algorithm is evaluated on several mixed data (i.e., data mixed with white noise) of various data length. The result shows that the MMFE algorithm can effectively reduce the deviation in entropy estimation as compared to that using MFE algorithm. The MMFE algorithm is further employed in the study to estimate the complexity and irregularity of vibration data of a roller element bearing for fault diagnosis. It is shown that the MMFE algorithm can effectively discriminate the four bearing operation conditions under study.

The performance of organic thin film transistors (OTFTs) with dual pentacene channel layer (DL) has been studied. Silvaco ATLAS is used to simulate the influence of different device structure parameters. The result shows that the variation in contact placement, thickness of high mobility pentacene layer, the total thickness of two pentacene layers and the height of source and drain result in different ON- and OFF-state current characteristics. The device structure having dual channel layers of pentacene with bottom source/drain contact has the better ON/OFF current ratio at a carefully chosen pentacene thickness and source/drain height.

With this expected future advance of HVDC, the use of gas insulated transmission lines (GIL) for dc application are getting increasingly interesting. For now, the problem of surface charge accumulation on gas-insulator interface is one of the critical factors for the development of DC-GIL. In many previous works, the characteristics of surface charges accumulation on insulator was investigated. However, the impact of temperature was neglected in most studies. In fact, a heating phenomenon is known to occur in transmission channel. Due to the surface charge accumulation is closed to the conductivity of insulator and insulated gas, and the conductivity of insulated materials are influenced deeply by temperature. Thus, this paper describes an experimental setup with the ability to simulate the heat transfer progress in DC-GIL, for measuring surface charge of insulator. It is hoped to study the impact of temperature on surface charge accumulation on insulator in future work.