Traffic flow prediction is a key challenge in intelligent transportation, and the ability to accurately forecast future traffic directly affects efficiency of urban transportation systems. However, existing deep learning-based models suffer from following issues: First, CNN- or RNN-based are limited by their architecture unsuitable for modeling long-term sequences. Second, most Transformer-based methods focus solely on data itself during embedding, neglecting implicit information behind...

Objectives To analyse differences in regional distribution and inequality health-resource allocation at the hospital primary health centre (PHC) levels Shanghai over 7 years. Design A longitudinal survey using 2010–2016 data, which were collected for analysis. Setting The study was conducted PHC Shanghai, China. Outcome measures Ten indicators used to measure levels. In addition, Theil Index calculated allocation. Results All quantities of healthcare resources per 1000 people hospitals PHCs...

While videoconferencing is prevalent, concurrent participation channels are limited. People experience challenges keeping up with the discussion, and misunderstanding frequently occurs. Through a formative study, we probed into design space of providing real-time transcripts as an extra communication for video meeting attendees. We then present MeetScript, system that provides parallel through interactive transcripts. MeetScript visualizes discussion chat-alike interface allows attendees to...

Photovoltaic power forecasting plays a crucial role in the integration of renewable energy into grid. However, existing methods suffer from issues such as cumulative multi-step prediction errors and limitations traditional evaluation metrics (e.g., MSE, MAE). To address these challenges, this study introduces DTCformer, generative model based on Autoformer. The proposed integrates Temporal Convolution Feedforward Network module Variable Selection Embedding module, effectively capturing...

A 2.5nm-thickness molybdenum diselenide (MoSe2) saturable absorber (SA) is prepared by electron beam evaporation (EBE) method. Applying the MoSe2 SA to an acousto-optic (AO) Q-switched fundamental laser, a dual-loss-modulated intra-cavity optical parametric oscillator (IOPO) has been experimentally realized. The signal-pulse train from this IOPO 0.0053 standard deviation (SD) of pulse amplitude. When applied IOPO, signal compressed maximum 68%, peak power increases 274%, and nonlinear...

Prioritized experience replay (PER) is an important technique in deep reinforcement learning (DRL). It improves the sampling efficiency of data various DRL algorithms and achieves great performance. PER uses temporal difference error (TD-error) to measure value experiences adjusts probability experiences. Although can sample valuable according TD-error, freshness also character implicitly reflects potential Fresh are produced by virtue current networks they more for updating network...

Despite the growing capabilities of short-term prediction photovoltaic power, we still face two challenges to longer time-range predictions: error accumulation and long-term time series feature extraction. In order improve range accuracy this paper proposes a seq2seq model TCNformer, which outperforms other state-of-the-art (SOTA) algorithms by introducing variable selection (VS), long- extraction (LSTFE), one-step temporal convolutional network (TCN) decoding. A VS module employs...

Tunnels water leakage detection in complex environments is difficult to detect the edge information due structural similarity between region of seepage and wet stains. In order address issue, this study proposes a model comprising multilevel transformer encoder an adaptive multitask decoder. The layered extract characteristics information, decoder comprises network branches. branches generate ground truths stains through threshold value transmit them for training. converged network, U-net,...

The waterproof curtain plays an important role in the dewatering of a deep foundation pit. Recognition depth inserted into confined aquifer at different depths may help control ground subsidence due to dewatering, but analysis interaction between and requires further study. In this study, we mainly analyze relationship based on shield shaft pit Qianhai-Nanshan tunnel project Shenzhen. Our numerical simulation results show that around decreases with increase aquifer, when completely...

A dense network of underground tunnels usually makes subsequent cross existing in the process urbanization. In this study, we mainly study deformation subway line 11 caused by under-crossing Qianhai–Nanshan Integrated Drainage Deep Tunnel Shenzhen. We analyzed displacement during tunnel according to monitoring data. Our results demonstrate that detection is an effective measure for controlling impact tunnels. The aim paper may provide guidelines some are under-crossed with similar conditions.

A dual-loss Q-switched laser was realized by applying a home-made 2.5 nm thick WTe2 to an acousto-optic modulated laser. Besides conventional pulse compression, the inclusion of could obviously improve average output power 29.6% and reduce threshold 20%, even with introduced 8% nonsaturable loss. The output-power improvement compression in this resulted 113.7% increase peak power. According measured spectra photoluminescence WTe2, laser-induced excitonic light emission may be reason for laser's gain.

Since small objects occupy less pixels in the image and are difficult to recognize. Small object detection has always been a research difficulty field of computer vision. Aiming at problems low sensitivity poor performance YOLOv3 for objects. AFYOLO, which is more sensitive was proposed this paper. Firstly, DenseNet module introduced into low-level layers backbone enhance transmission ability information. At same time, new mechanism combining channel attention spatial improve feature...

Moving on natural muddy terrains, where soil composition and water content vary significantly, is complex challenging. To understand how mud properties robot-mud interaction strategies affect locomotion performance mud, we study the terrestrial of a mudskipper-inspired robot synthetic with precisely-controlled ratios sand, clay, water. We observed non-monotonic dependence speed content. Robot was largest intermediate levels (25%–26%), but decreased significantly higher or lower Measurements...

This study introduces user experience into the research and design of virtual exhibition platform, through eye tracking experiment survey research, illustrates experience's influence on platform. Proposes platform based experience, opens up a broader space for design, provides rational point view designers to better grasp reference design.

This paper discusses the application of audio signal parameters extraction and analysis technology in multimedia vocal music teaching, studies teaching based on voice recognition. In process recognition, MFCC speech frame energy constitute a combination feature as parameters, improved DTW is taken pattern matching algorithm evaluation criteria. Matlab used to simulate system determine reliability its performance, running test realized successfully. The experimental results show that realizes...

At present, most of the methods used in research image semantic segmentation ignore low-level feature information image, such as space, edge, etc., which leads to problems that edge and small part is not precise enough accuracy result high. To solve this problem, paper proposes a multi-source fusion model generative adversarial networks based on FCN: SCAGAN. In VGG19 network, add super-pixel detection algorithm, introduce efficient spatial pyramid module reduce number parameters while adding...

Wind farm group integrated system consists of many Large-Scale a single wind farm, which will have total capacity 1,000MW (and above). Reactive- power fluctuation seriously worse the voltage in limited area nearby .However, Reactive optimization is to regulation reactive mostly, that cannot satisfy demand regulating for local system. Using level PCC and substation as target function, this paper applies an efficient method based on genetic algorithm extract compensation point optimize...

A method has been developed for the determination of mercury in cosmetics by suspension-sampling hydride generation atomic fluorescence spectrometry. About 500 mg cosmetics, was mixed with absolute ethanol and agar solution. The samples were agitated a vortex 10 min. procedure permits use direct calibration several cosmetic samples. influences instrument operating conditions, suspension preparation interferences on signals accuracy precision determined. results reference sample GBW09303a...