Search

Many electrical equipment malfunction text messages are collected during power system operation and maintenance procedures. These texts usually contain crucial information for maintenance and condition monitoring. Because these power system malfunction texts are characterized by multidomain vocabularies, complex-syntactic structures, and long sentences, it is challenging to for automated systems to capture their semantic meaning and essential information. To address this issue, we propose a hybrid natural language processing (hybrid-NLP) algorithm to extract entities that represent electrical equipment. This algorithm is composed of a dictionary-based method, a language technology platform (LTP) tool, and the bidirectional encoder representations from a transformers-conditional random field (BERT-CRF) model. Significantly, the softmax output layer of the bidirectional encoder representations from the transformers (BERT) model is replaced by the conditional random field (CRF) algorithm to strengthen the contextual relationships between words and thus solve the local optimization of the preferred word label. The effectiveness of the proposed hybrid-NLP method is verified on a realistic dataset. Moreover, a statistical analysis is conducted to provide a reference for the operation and maintenance of power systems.

We introduce a novel concept for the design of functional surfaces of materials: Spatial surface charge engineering. We exploit the concept for an all-solid-state, epitaxial InN/InGaN-on-Si reference electrode to replace the inconvenient liquid-filled reference electrodes, such as Ag/AgCl. Reference electrodes are universal components of electrochemical sensors, ubiquitous in electrochemistry to set a constant potential. For subtle interrelation of structure design, surface morphology and the unique surface charge properties of InGaN, the reference electrode has less than 10 mV/decade sensitivity over a wide concentration range, evaluated for KCl aqueous solutions and less than 2 mV/hour long-time drift over 12 hours. Key is a nanoscale charge balanced surface for the right InGaN composition, InN amount and InGaN surface morphology, depending on growth conditions and layer thickness, which is underpinned by the surface potential measured by Kelvin probe force microscopy. When paired with the InN/InGaN quantum dot sensing electrode with super-Nernstian sensitivity, where only structure design and surface morphology are changed, this completes an all-InGaN-based electrochemical sensor with unprecedented performance.

The heterostructuring and doping concepts have proved to obtain a novel n-InGaN/p-Cu2O nanowire (NW) photoanode by strong enhancement of the photocurrent compared to a bare InGaN NW photoanode in solar water splitting. The large photocurrent is due to the maximized photocarrier separation and hole transfer to the surface in the depletion zone of the p–n heterojunction established by the p-Cu2O layer, forming a thin, uniform shell-layer around the n-InGaN NW core by electrodeposition. For sufficiently thin Cu2O layers, the upward energy band bending in the depletion zone extends up to the surface for optimized hole transport and surface reaction. Thick Cu2O layers on top of the InGaN NWs act as common photocathodes. The functional InGaN/Cu2O heterostructure core-shell NW photoanode is chemically self-stabilized at positive applied voltage by a thin CuO surface layer. Final deposition of the earth-abundant NiOOH co-catalyst boosts the photocurrent of the InGaN/Cu2O/NiOOH complete NW photoanode into the competitive mA/cm2 range.

HfO2 and SiO2 optical thin films were etched using ion beam etching (IBE), reactive ion etching (RIE) and inductively coupled plasma etching (ICPE) techniques. The influence of reactive gas and ion bombardment energy on the etching rates and surface morphologies of these two coatings was comparatively studied. Low ion bombardment energy with Ar/CHF3 (40/20sccm) realized a preferentially etching of SiO2 with a high rate of 50nm/min, and it almost had no influence of HfO2 coatings.

In recent years, the automobile industry has developed rapidly and the drivers’ unregulated driving behaviors have caused a lot of traffic safety problems. By using the topic model, this paper can find the relationship between driving pattern with driving behavior and provide technical support for traffic safety. Taking the shortage of the current mainstream topic models pLSA and LDA (Latent Dirichlet Allocation) into account, this paper proposes an improved LDA model with time labels, namely the T-LDA model to identify driving patterns. The experimental results show that the improved model can effectively extract a series of continuous driving behavior characteristic and improve the accuracy of driving pattern recognition.

Broadband multilayer dielectric gratings (MDGs) with rectangular HfO2 grating profile were realized for the first time using a novel fabrication process that combines laser interference lithography, nanoimprint, atomic layer deposition and reactive ion-beam etching. The laser-induced damage initiating at the grating ridge was mitigated for two reasons. First, the rectangular grating profile exhibits the minimum electric-field intensity (EFI) enhancement inside the grating pillar compared to other trapezoidal profiles. Second, our etching process did not create nano-absorbing defects at the edge of the HfO2 grating where the peak EFI locates, which is unavoidable in traditional fabrication process. The fabricated MDGs showed a high laser induced damage threshold of 0.59J/cm2 for a Ti-sapphire laser with pulse width of 40 fs and an excellent broadband diffraction spectrum with 95% efficiency over 150 nm in TE polarization.

We demonstrate an all-solid-state potentiometric sensor constructed from solid-state InN/InGaN sensing- and reference electrodes with the epitaxial InN/InGaN layers directly grown on Si substrates. The sensor, evaluated in KCl aqueous solution, exhibits super-Nernstian sensitivity of −78 mV/decade with good linearity for concentrations of 0.01–1 M, which is the physiologically relevant range. Good stability and re-usability are demonstrated by a long-time drift below 0.2 mV h−1 and standard deviation of 8 mV for repeated measurements over 10 d. These properties fulfil the requirements for compact, robust and highly sensitive all-solid-state sensors and sensor arrays.

A comparative study was performed to investigate how etching methods and parameters affect the properties of S i O 2 and H f O 2 coatings. S i O 2 and H f O 2 single layers were prepared by electron-beam evaporation (EBE), ion-beam assisted deposition (IAD), and ion-beam sputtering (IBS). Then, ion-beam etching (IBE), reactive ion etching (RIE), and inductively coupled plasma etching (ICPE) were used to study the influence of ion bombardment energy and chemical reaction on the etching rates and properties of the prepared S i O 2 and H f O 2 single layers. For S i O 2 coatings, chemical reaction plays a dominant role in determining the etching rates, so ICPE that has the strongest C H F 3 plasma shows the highest etching rate. Moreover, all three etching methods have barely any influence on the properties of S i O 2 coatings. For H f O 2 coatings, the etching rates are more dependent on the ion bombardment energy, although the chemical reaction using C H F 3 plasma also helps to increase the etching rates to some extent. To our surprise, the ion bombardment with energy as high as 900 V does not change the amorphous microstructure or crystalline states of prepared H f O 2 coatings. However, the high-energy ion bombardment in IBE significantly increases the absorption of the H f O 2 coatings prepared by all deposition techniques and decreases their laser damage resistance to different extents.

For the sol-gel method, it is still challenging to achieve excellent spectral performance when preparing antireflection (AR) coating by this way. The difficulty lies in controlling the film thickness accurately. To correct the thickness error of sol-gel coating, a hybrid approach that combined conventional sol-gel process with ion-beam etching technology was proposed in this work. The etching rate was carefully adjusted and calibrated to a relatively low value for removing the redundant material. Using atomic force microscope (AFM), it has been demonstrated that film surface morphology will not be changed in this process. After correcting the thickness error, an AR coating working at 1064 nm was prepared with transmittance higher than 99.5%.

Multimedia applications make researchers focus on the simultaneous effect of visual and auditory stimulation. Previous research revealed many facts about visual-auditory interaction. But the relationship between colors and pitch discrimination was hardly discussed. In this paper, the influence of colors on pitch discrimination of pure tones was investigated. Different color stimuli were used in subjective experiments of measuring the accuracy of pitch discrimination, and the data were compared with auditory-only experiments. Experimental results showed that the accuracy of pitch identification with color stimuli increases. Warm colors and cold colors did not have remarkable difference. The accuracy increased most when the color was blue. Further analysis showed that if the discrimination task was more difficult,the influence of colors was correspondingly more observable.

Photon detector is one of the key units in the integrated optical devices. Due to the weak atom-photon interaction, the detection efficiency of a single photon by a bare atom (as a photon detector aside the waveguide) is relatively low (the most reachable value is shown as $50%$, which has being approached experimentally). In order to overcome such a limit, a pair of spatially separated two-level atoms (the photon lenses) with adjustable inflection and transmission coefficients of the photon are introduced to generate an atomic cavity, in which the detected probability of the waveguide photon by the single detector can be significantly enhanced (even reach to $100%$ under certain conditions). The proposal is demonstrated specifically by a circuit QED system for highly effective detection of a single microwave photon.

Classification of musical instrument is resulted from musicology. Thus, automatic identification of instrument families not only benefits the study in musicology, but also worth of attention in MIR. Based on a database consists of 2177 clips from Chinese and Western instrumental music, the experiments provided in this paper evaluate the ability of automatic Chinese and Western instrument identification by confusion matrices. Regarding various features and classifiers, the experiment results revealed the similarity and difference between Chinese and western instrumental music, which produce new possibilities for the study in identifying and classifying Chinese and Western instrument families.

Instrumental music is often classified or retrieved in terms of instruments played in it. With a large database consists of Chinese traditional music and western classical music, this paper extracted several features to automatically classify Chinese and western instruments by SVM classifier, and analyzed the classification results.

According to numbers of music cognitive experiments, moods or emotions in music could be categorical. Since mood classifications are commonly used to structure the large collections of music available on the Web, automatic discrimination between mood taxonomy of Chinese traditional music and Western classical music would be a valuable addition to music information retrieval (MIR) systems. In this paper, three content feature sets are extracted directly from the waveform audio clips, and then two mood taxonomy models are implemented. A Bayesian network is trained to classify the discrete mood categories. Finally, because the already-known algorithms have rarely applied to the Chinese traditional music, the comparative experimental result between Chinese and Western music evokes further research necessities.

In this work, methods of design and fabrication of photonic bandgap structures (PBG) and photonic crystal waveguides based on SOI (silicon on insulator) are presented. In theory, a method that incorporates the plane wave expansion (PWE) method based on supercell with the finite-difference time-domain (FDTD) method with a perfectly matched layer (PML) boundary condition has been investigated. At first, PWE simulation will present a band structure. Then according to the band structure, FDTD tool can simulate a light propagation and can obtain optimized parameters easily. With the method, several photonic crystal devices suitable for 248nm Deep UV lithography and 0.18um ion-beam etching are designed and fabricated.