Search

Sentence alignment provides multi-lingual or cross-lingual natural language processing (NLP) applications with high-quality parallel sentence pairs. Normally, an aligned sentence pair contains multiple aligned words, which intuitively play different roles during sentence alignment. Inspired by this intuition, we propose to deal with the problem of sentence alignment by exploring the semantic interactionship among fine-grained word pairs within the framework of neural network. In particular, we first employ various relevance measures to capture various kinds of semantic interactions among word pairs by using a word-pair relevance network, and then model their importance by using a multi-view attention network. Experimental results on both monotonic and non-monotonic bitexts show that our proposed approach significantly improves the performance of sentence alignment.

The audit of waste materials can reduce the loss of state assets and strengthen the construction of clean government. With the construction of large data platform in power grid, the generation of massive data has provided convenience for audit supervision, but the identification and extraction of key information has also brought difficulties to the traditional audit work. Fistly,the article classified the waste materials of electricity, and used data mining technology to carry out important data association and processing for the equipment after classification, got the state information before abandonment, and made reasonable audit for scrap. Then, got the information of volume and weight under normal condition, compared with the scrapped, carried out the audit of “accounts and things” consistency after scrapping. The engineering practice shows that the use of digital audit can effectively improve audit efficiency and discover problems in time, and is worthy of further popularization and application.

In order for practical applications in vacuum electron devices, the robustness of scandia-doped dispenser (SDD) cathode is a major concern. Emission properties after air exposure and gas poisoning have, therefore, been studied. A full restoring of emission has been proved after the cathodes were exposed to air and reactivated for a time much shorter than initial activation. Poisoning experiments demonstrated that the poisoning pressure thresholds of residual gases for the SDD cathode are about half to one order of magnitude higher at equivalent temperatures but with higher emission level in comparison with that of an excellent M cathode. The cathode is easier to recover after poisoning and can operate stably in a nonideal environment. These features are thought to be related to the surface Ba–Sc–O structure which has closely chemical association, better stability for air exposure, relatively high tolerance to gas poisoning, and the semiconductor emission characteristic.

Near infrared spectroscopy combined with chemometrics and pattern recognition has become a primary focus in the discriminant analysis of agricultural products. To date, most studies have focused on using a single classifier to discriminate the origins, varieties and grades of products. Others have focused on using multiple classifier fusion by weighted voting. Due to their attributes of continuity and internal similarity, discriminant models sometimes present poor performance. In this study, we achieved better performance by applying multiple classifier fusion models, including support vector machine (SVM), discriminant partial least squares (DPLS) and principal component and Fisher criterion (PPF). PPF showed continuity and similarity among different parts of tobacco leaves [i.e. upper (B), cutter (C) and lug (X)]. The similarities between each class and the others were quantified to values, and the sum of the similarity values of each class was defined as its similarity. SVM–DPLS–PPF fusion by voting and similarity constraint for decision resulted in better performance, with the correct discriminant rate improved on average by 14.1%, 8.2%, 17.3% and 4.6% compared with those achieved using SVM, DPLS, PPF and SVM–DPLS–PPF fusion by weighted voting for decision, respectively; in addition, the incorrect discriminant rate between B and X was reduced to zero. Therefore, we demonstrated the feasibility of using SVM–DPLS–PPF fusion by voting and similarity constraint for decision to discriminate between different parts of tobacco leaves. This technique could provide a new method for tobacco quality management, computer-aided grading and intelligent acquisition. It also provides a new discriminant method for analysing the attributes of continuity and similarity of agricultural products using near infrared spectroscopy.

2 µm, 3-5 µm and 8-12 µm infrared lasers are located in the atmospheric transmission window, which have broad applications in laser medicine, laser imaging, environmental monitoring, lidar, chemical remote sensing and infrared countermeasure. Based on the optical nonlinear frequency conversion technology and nonlinear optical crystals, it has obvious advantages in achieving middle-long-wave infrared solid lasers, such as compact and simple structure, wide tunable wavelength range and high output power. Using ZnGeP2 crystal with 2 µm Ho single-doped solid laser as pump especially has an outstanding performance in middle-long-wave infrared field. In the aspect of average output power, it has reached the level of 102 W@3-5 µm, 12.6 W@8.2 µm and 3.5 W@9.8 µm. Moreover, they all have a beam quality factor M2 less than 3 and the corresponding optical-to-optical conversion efficiency of 3-5 µm is about 60%. This paper reviewed the research progress of 2 µm Ho single-doped solid laser and application of ZnGeP2 on middle-long-wave infrared in detail.

The Large Hadron Collider (LHC) is the world largest and most powerful accelerator for high energy physics. The High Luminosity LHC project (HL- LHC) foresees the construction and installation of important upgrades to increase the efficiency of the LHC machine. Hollow Electron Lenses (HEL) are one of the most promising methods being considered to control the beam halo and to improve the beam collimation system for HL-LHC. The source of the HEL is an annular shaped cathode which delivers a total current up to 5A. We developed a nano-sized Scandia doped dispenser cathode much smaller in size than the previously used B-type dispenser cathodes. The tests demonstrated that this cathode fulfils the specifications and can supply 5A full Space Charge Limited (SCL) current at a true temperature of about 850°C.

Temperature rise is one of the main factors that affect the safe operation of electric vehicle. In order to improve the heat efficiency of the motor, the water-cooled motor of circular circumferential waterway for mini electric vehicle was studied as an object in this research, the structural parameters of the channel are optimized and analyzed from the following aspects: the width of the adjacent channel, the width and the height of the channel. A numerical model for coupling the fluid field and temperature field of a motor with different waterway parameters is established, comparing and analyzing the temperature rise distribution of different cooling structure parameters, the optimum cooling system parameters are obtained, which provides a reference for the design of the cooling system of the small and medium sized water cooling motor.

With the increasing penetration of wind power, impacts of the wind power random nature on the power quality, especially on voltage quality, increase gradually. Flywheel energy storage system (FESS) has the ability of independent regulating active power and reactive power. This paper designed the energy management system (EMS) model of FESS which can be used to improve the power quality of the grid. The model is based on the bi level decision model. The upper layer uses the fuzzy algorithm to solve the charging and discharging power of the FESS through assess the state of FESS and the amount of wind power fluctuations, the lower layer control the energy exchange between FESS and grid through by control the back-to-back converters based on dual-loop control. The voltage fluctuation of the PCC was compared using the FESS before and after in the MATLAB/Simulink simulation environment. The simulation results verify the validity of the model. The model can improve the utilization efficiency of FESS and the power quality of the system through the cooperation between the wind power and energy storage system.

Based on two kinds of maximum power point tracking (MPPT), this paper proposed an improved variable step size incremental conductance algorithm which combined with constant voltage tracking. The operating point of PV array is located near the maximum power point (MPP) rapidly by using the fixed voltage method, and the PV array is controlled to run at the MPP quickly and accurately adopting different disturbance step. Grid-connected inverter adopts the double closed-loop control strategy of based on feed-forward decoupling to realize the decoupling control of active and reactive power. In PSCAD/EMTDC, the model of grid-connected photovoltaic system (GCPS) has been implemented and verified the effectiveness of the improved MPPT algorithm.

Interfacial microstructures of thermosonic Au wire bonding to an Al pad of die were investigated firstly by high-resolution transmission electron microscopy (HRTEM) and X-ray micro-diffractometer. The equal-thickness interference structures were observed by HRTEM due to diffusion and reaction activated by ultrasonic and thermal at the Au/Al bond interface. And X-ray diffraction results showed that three different interplanar crystal spacings (‘d’ value) of the interfacial microstructures were 2.2257 A, 2.2645 A, and 2.1806 A respectively from the high intensity of diffraction to the low intensity of diffraction. These indicated that the intermetallic phase AlAu2 formed within a very short time. It would be helpful to further research wire bonding technology.

For large-scale wind power access to the grid based on thermal power peaking, the structure of power and peaking capability restrict the accessing of large-scale wind power. Large-scale energy storage system can achieve load space-time translation and improve security and economy of power grid operation. In this paper, considering factors of storage cost, wind power investment cost and wind power revenue, a storage capacity optimal allocation method by using large-scale storage is presented to improve the grid ability of integrating wind power. This method is applied in simulations where a typical power load of a province in 2010 is considered. The results verify the effectiveness and feasibility of the proposed method.

The stable bonding of gold wire is becoming more and more important with the development of high integration and miniaturization of semiconductor. At the beginning of the thermosonic wire bonding cycle, the auto wire bonder uses an Electronic Flame-Off (EFO) unit to create a Free Air Ball(FAB). As FAB affects the quality of the first bond directly, consistency of FAB formation is very important for the stable wire bonding. In this study, a high-speed camera system was used to record the EFO and FAB formation process under different EFO currents. Continuous image sequences of FAB formation were processed by using matlab software. Finally, the influence law of EFO current on FAB fomation was obtained.

A dynamic model of back-to-back voltage source converter under synchronous rotating d-q coordinate system is developed. The system is transformed to a new equivalent nonlinear system which consists of a linear decoupled subsystem and a nonlinear subsystem characterizing the dynamics of the dc-link voltage in terms of the state variables in the linear subsystem via the feedback linearization technique. Then a decoupled controller which consists of an outer loop for determination of d- and q-axis components of reference currents and an inner loop for tracking the reference currents is designed for active and reactive power exchange control between the VSC and the two ac systems. The validity of the proposed controller is demonstrated by the simulation results under electromagnetic software PSCAD/EMTDC and experimental results on a 30 kVA/380 V prototype system, respectively.

Ultrasonic features in bonding area are of interests for researchers in the field of microelectronics packaging. In this study, the interface characteristics of bonding were detected by using XD7100 X-ray, the velocity of ultrasonic vibration at bonding tool was determined with a PSV-400-M2 high frequency (1.5 MHz) laser Doppler vibrometer, micro- characteristics of the specimens of thermosonic flip chip bonding were inspected by using a high resolution transmission electron microscope. Results show that bubbles at the interfaces of reflow soldering flip chip were observed, however they were not discovered during thermosonic flip chip, the interface characteristics of thermosonic flip chip is better than that of reflow soldering flip chip. The acceleration of the ultrasonic vibration was about 500000 m/s2 for a 62.73 kHz bonding frequency. This strong mechanical effect activates dislocations in the metal crystalline lattice and enhances atomic diffusion across the bonding interfaces, and these differ from melting mechanism of reflow soldering.

Cross-section features in a flip-chip bonding area were inspected by using a high resolution transmission electron microscope (HRTEM). For the better flip chip bonding variables, depth of atom diffusion at Au/Al interface of ultrasonic flip chip (FC) bonding was about 500 nm by using a line scanning of HRTEM, and depth of atom diffusion at Au/Ag interface of ultrasonic FC bonding was about 200 nm. If Au/Ag interface received ultrasonic energy first by changing the process of ultrasonic FC, depth of Au/Ag atomic diffusion was about 450 nm, and depth of Au/Al atomic diffusion was about 250 nm. An ultrasonic energy conversion law was discovered that the ratio of first interface to next interface along ultrasonic energy transmitting path in ultrasonic FC bonding was about 2.28:1, which indicated an optimizing process of ultrasonic FC bonding, and a new way of FC bonding was designed for the first interface along the path of ultrasonic energy transmitting.

LED solid state lighting (SSL) has a good prospects, but there are some bottlenecks, the one is that efficiency of light emitting is not enough for general lighting, the other is too much heat generated by higher power which affect the efficiency and life of LED. Flipchip's applying is good for improvement of efficiency of light emitting of LED and especially for thermal management. The effect of the layout and the number of bumps of LED flipchip to thermal management of LED light was analyzed by using FEM simulation. The results show that better effect of thermal management can get after optimizing layout of bumps, and also show that the more bump number used the better thermal management can get, but there is a best optimal number as there are some other limiting factors such as structure and cost

In this paper, we propose a method to predict ultra-short term wind power production with chaotic time series. Through analysis of the data measured from real wind farm, the ultra-short term prediction model of wind power is established based on the space theory of chaotic time series phase. The wind power of a real wind farm in northeast China in some future time including 15 m, 30 m and 1 h are predicted by the prediction model finally. Test results are presented in the paper.