Search

Recently, ensemble learning paradigm has shown great potential to achieve better prediction performance in the hyperspectral image classification. However, in the traditional methods, each classifier independently searches for the optimal spectral feature subspace in series and some important features are searched repeatedly, which leads to high computing redundancy and low effective utilization of features. In this article, an evolutionary multitask ensemble learning model (EMT_EL) for hyperspectral image classification is designed. First, the model formulates the spectral feature subspaces generation into a multitask optimization problem to concurrently search for optimal feature subspaces for multiple classifiers, which would be capable to select more informative and representative feature subspaces effectively. Second, seeking the optimal feature subspace for one base classifier can assist in the optima-seeking process for some other base classifiers via sharing the useful features, which can accelerate converge toward the direction of the optimal feature subspace, avoid trapping in local optimal subspace and improve searching capability. Third, randomization-enhanced genetic operators are designed for effective and reasonable feature selection, which can facilitate the exchange of information and improve the joint searching efficiency of the feature subspace. Eventually, the quality of generated spectral feature subspaces for each base classifier is improved and the feature sharing can parse HSI data by knowing which spectral features are important. Experimental results demonstrate that the proposed method can generate the appropriate feature subspace for each base classifier, thus it has outstanding classification performance on the different hyperspectral datasets.

Dielectric barrier discharge excitated by pulsed power is a promising approach for producing nonthermal plasma at atmospheric pressure, but pulsed power generators vary widely in performance and should be chosen according to application requirements. In this paper, a repetitive unipolar nanosecond-pulse generator is constructed using resonant charging and one-stage magnetic compression circuits, where IGBT and magnetic switches are the key units, respectively. The generator is capable of providing repetitive pulses with a voltage of up to 30 kV and duration of 70 ns at a 300 Ω resistive load. Output pulse voltage can be adjusted by varying ac input voltage or trigger pulse-width. This compact and convenient generator has been used successfully to produce stable dielectric barrier discharge. [ABSTRACT FROM AUTHOR]

Dielectric barrier discharge (DBD) is proved to be an effective method for surface treatment of polymers over the past 20 years and is now widely used in the fields of surface modification. In this paper, surface treatment of polyethylene terephthalate (PET) films for improving the hydrophilicity using DBD excited by unipolar nanosecond pulses is presented. Two typical discharges exhibiting homogeneous and filamentary modes are obtained under certain experimental conditions, and then, they are used to modify the surface of PET films. The microstructure and properties of the film surface before and after treatment are characterized with water contact angle measurement, scanning electron microscope, atomic force microscope, and X-ray photoelectron spectroscope. The results show that, with the treatment power density of 192 and 158 mW/cm2 in filamentary and homogeneous modes, respectively, surface morphology and property after treatment are significantly changed, and the physical etching and introduction of oxygen-containing polar functional groups account for the decease of surface contact angle and the increase of surface roughness. Compared with the filamentary DBD treatment, the homogeneous DBD (HDBD) is more effective in surface treatment, such as lower contact angle, higher roughness, and more oxygencontaining function groups after treatment. It is attributed to the fact that the HDBD can achieve a uniform treatment and canmake the physical and chemical interaction much sufficient. [ABSTRACT FROM AUTHOR]

Nanosecond-pulse breakdown indicates special characteristics depending on the pulse rise-time and duration. Based on a repetitive nanosecond-pulse generator, breakdown phenomena of parallel-plane gaps in nitrogen were investigated with single pulse and repetitive bursts under different gap conditions. The relationships between applied voltage, pulse repetition frequency, breakdown time lag, repetitive pulse stress time and the number of applied pulses are presented. The curves regarding E-field strength, breakdown time lag and gas pressure are also obtained. The experimental results show that breakdown characteristics with repetitive nanosecond-pulses are different from that with single pulse. Repetitive nanosecond-pulse breakdown is concerned with the accumulation effect that is attributed to residual ions and metastable species survived from previous pulses. in addition, a modified empirical formula about E-field strength, breakdown time lag and gas density is given for the breakdown data. [ABSTRACT FROM AUTHOR]

Based on a Marx generator and a coaxial pulse forming line, an experimental investigation of surface flashover characteristics in vacuum is conducted by using nanosecond pulses of 10 ns rise time and 30 us full width at half maximum (FWHM). Insulator dielectrics chosen for this investigation are Teflon, PMMA and Nylon. The tested factors include gas pressure, cone angle of conical frustum, diameter and length of cylindrical insulator, material and shape of electrode, and contact style between insulator and electrodes. The effects of these parameters on the surface flashover characteristics are described and analyzed in this paper. In addition, the character of flashover time lag in the nanosecond range, and surface flashover theory in vacuum charged by nanosecond pulses are also discussed. [ABSTRACT FROM AUTHOR]

Repetitive pulsed power is becoming an important area of high-power technology. Dielectric failure data concerning electrical insulation play a basic role, but breakdown has been inadequately studied for the repetitive nanosecond-pulse conditions. This paper is concerned with the breakdown characteristics of tip-plane gas gaps under repetitive burst conditions at variant repetition rates (rep-rates) and diverse gap distances. The relationship among applied voltage, breakdown time lag, number of applied pulses to breakdown, repetitive pulse stress time, and rep-rates is presented. The experimental results presented show that breakdown polarity dependence is not distinct. The data also indicate that significant concentrations of excited particles and residual charges would be formed during the consecutive nanosecond pulses and would present a memory effect that affects the development of gas breakdown. In addition, the detachment of negative ions, cathode collision of positive ions, and deexcitation of metastable species can provide the source of avalanche-initiating electrons. [ABSTRACT FROM AUTHOR]

This paper illustrates images of dielectric-barrier discharge (DBD) excitated by repetitive unipolar nanosecond pulses in atmospheric air. Different images are obtained for three barrier dielectrics and various gap lengths. Discharge images show that the nanosecond-pulse DBD observed with the eye is a diffuse glowlike mode when air gap is 2 mm; clear filamentary discharge is seen with the eye in the case of air gap of 8 mm. The results show that gap length and dielectrics are the important factors influencing the transition from diffusive to filamentary discharges. [ABSTRACT FROM AUTHOR]