Abstract:During extreme rainfall events, urban roads can function as surface runoff conduits when integrated with drainage systems, providing a cost-effective flood mitigation solution. However, the lack of reliable computational tools for modeling flow diversion at road intersections makes it challenging to accurately estimate drainage flow, limiting broader application of this technique. This study focuses on T-shaped intersections in urban road networks and conducts both hydraulic experiments and computational fluid dynamics (CFD) simulations under three distinct downstream boundary conditions: free outflow, backwater at the main road end, and backwater at both road ends. Based on the obtained data, a high-precision, low order flow diversion model for intersections is developed, with a calculation error basically within ±15%. The proposed model provides a simple and practical tool for engineering applications and can be seamlessly integrated into the widely used one-dimensional urban hydrological model SWMM (storm water management model). This integration significantly enhances the accuracy of flow distribution calculations at intersection nodes, enabling more precise design of the large-scale road major drainage system.

Abstract:To investigate the distribution characteristics of wind parameters near shorelines, a two-dimensional geometric model is developed using computational fluid dynamics (CFD). The study employed the SST k-ω turbulence model and a multiphase flow model to simulate wind parameter distributions under the influence of two-dimensional wind over calm water. The effects of fetch distance, water depth, incoming wind speed, and underwater terrain slope on the distribution characteristics were analyzed. Results indicate that for fetch distances belowe 60 m, wind speed near the water surface exceeds the inlet wind speed, demonstrating an acceleration effect. Beyond 60 m, wind speed increases with height until stabilizing at a specific elevation. Terrain slope variations exhibit negligible effects on wind parameter distribution, while water depth shows strong correlations. Deeper water leads to a nonlinear increase in gradient wind height and significantly alters wind speed profiles. Although inlet wind speed does not affect gradient wind height or wind structure, it does impact the near-surface acceleration effect.

Abstract:The mechanical behavior of transmission towers and the maximum jump height of conductors following ice-shedding are critical factors in tower head design. In ultra-heavy ice zones, ice thickness on ultra-high voltage direct current (UHV DC) line can reach 60 mm to 80 mm, exceeding the maximum values specified in current transmission line design codes. This study establishes finite element models of UHV DC tower-line systems in ultra-heavy ice zones and numerically simulates their dynamic responses under ice-shedding conditions for varying span lengths. The analysis evaluates tower stresses, longitudinal unbalanced tensions, and maximum conductor jump heights to assess both structural performance and electrical isolation clearances. Results indicate that longitudinal unbalanced tensions surpass estimates from current design codes, and maximum conductor jump heights exceed predictions from existing empirical formulas. To enhance design accuracy, the study proposes revised values for longitudinal unbalanced tensions and modifications to the conductor jump height formula.

Abstract:The Tarim river’s main stream contains substantial deposits of aeolian sand. Due to the scouring effect of seasonal floods, sliding failures frequently occurs along the riverbanks. To investigate the failure mechanisms, we conducted indoor direct shear, compression, and penetration tests to explore the variation in the mechanical properties of aeolian sand under different water content and dry density conditions. The results show that as water content increases, cohesion initially increases and then decreases, reaching a maximum at the optimum moisture content. This relationship can be expressed by a quadratic function, whereas the internal friction angle decreases linearly. The formation of a viscous water film on the particle surfaces contributes to these effects. Beyond the optimal water content, the viscosity of the water film weakens, resulting in a decline in cohesion and increased sliding between particles. The thickened water film also reduces sliding friction as particles roll over one another. As dry density increases, both cohesion and internal friction angle increase linearly. This is due to decreased particle spacing, enhanced van der Waals forces, and improved inter-particle locking. These factors collectively lead to greater resistance to shear displacement and higher internal friction. Additionally, with increasing water content, both the compression coefficient and modulus of resilience show a linear increasing trend. Under the same axial stress, higher water content leads to a thicker water film, reduced interparticle resistance during displacement, greater compressibility, and higher rebound potential. Conversely, increasing dry density results in a linear decrease in the compression coefficient and a linear increase in the modulus of resilience. Closer particle contact and increased resistance during displacement contributes to reduced compression deformation and enhanced elastic rebound. The permeability coefficient also decreases linearly with increasing dry density, ranging from 1×10^-4 cm/s to 3×10^-4 cm/s, which is 2 to 3 orders of magnitude lower than traditional theoretical estimates. A modified theoretical formula for calculating the permeability coefficient is proposed. After eliminating the errors caused by the low dry density, the experimental values closely match the empirical calculations, with the relationship described by a linear function. As dry density increases, the resistance to water molecule migration through soil pores rises, resulting in decreased permeability.

Abstract:Taking the Haerwusu open-pit coal mine as a case study, this research addresses the challenge of achieving a high design capacity of 35 Mt/a under the constraints of shortened working line length and reduced mining district width, following the retraction of the first mining district’s working line into the mining license boundary. Based on principles of technical feasibility and economic rationality, and with considering the influence of working line length on production stripping ratio, stripping haulage distance, and raw coal production capacity, a mathematical model was established to minimize the annual total stripping cost. The economically optimal working line length was determined to range from 1 620 m to 2 315 m. Building upon this, and with integrating the production capacity and equipment configuration of existing mining equipment at Haerwusu, it was determined that the average annual advancing speed of the working line should be maintained between 400 m to 515.25 m. A working line length of 1 820 m was found to maximize the operational efficiency of the mining equipment. Based on the feasible range of advancing speeds, the Monte Carlo method was innovatively used to optimize the working line shape. Three working line layout and development schemes, all satisfying the constraints of the working line length and annual advancing speed, were proposed. To evaluate these schemes, a CRITIC-TOPSIS comprehensive evaluation model based on objective weighting was constructed, incorporating six key indicators: average production stripping ratio, weighted average haulage distance and lifting height for both stripping and raw coal, and the average maximum advancing speed of the working line. Evaluation results show that scheme 2 exhibits the closest proximity to the ideal solution. This scheme enables Haerwusu open-pit coal mine to achieve its 35 Mt/a raw coal production target within a narrow mining district, while delivering optimal technical and economic outcomes.

Abstract:Although few-shot action recognition based on the metric learning paradigm has achieved significant success, it fails to address the following issues: 1) inadequate action relation modeling and underutilization of multi-modal information; 2) challenges in handling video matching problems with different lengths and speeds, and misaligned video sub-actions. To address these limitations, we propose a two-stream joint matching (TSJM) method based on mutual information, which consists of two modules: multi-modal contrastive learning module (MCL) and joint matching module (JMM). The MCL extensively explores inter-modal mutual information relationships, and thoroughly extracts modal information to enhance the modeling of action relationships. The JMM is primarily designed to simultaneously solve the aforementioned video matching problems. By integrating dynamic time warping (DTW) and bipartite graph matching, it optimizes the matching process to generate the final alignment results, thereby achieving high few-shot action recognition accuracy. We evaluate the proposed method on two widely used few-shot action recognition datasets (SSV2 and Kinetics), and conduct comprehensive ablation experiments to substantiate the efficacy of our approach.

Abstract:To address challenges in multi-time-step health prediction for lifting machinery, such as short data spans, high-frequency measurements, multi-dimensional feature complexity, and limited labeled data, this paper proposes a hybrid method combining convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) networks with an encoder-decoder architecture (ED-BLSTM). The method begins by chronologically organizing monitoring data, followed by segmenting and reconstructing the dataset while maintaining consistent input-output time step sizes. The processed data is first fed into a CNN to extract the main features, generating a multi-dimensional feature matrix. This matrix then trains a BiLSTM network within an encoder-decoder framework to build a predictive model for multistep forecasting of machinery health status. Comparative experimental results show that the method reduces validation loss by 0.097% to 0.474% and prediction loss by 1.230% to 1.411%, outperforming current mainstream approaches. These results demonstrate its potential to advance predictive maintenance in industrial equipment.

Abstract:Cybersecurity situational awareness technology plays a critical role in assessing network security status, predicting potential attack paths, and assisting administrators in implementing effective defenses. Traditional methods for network situation assessment mostly rely on theoretical analysis, limiting their practicality in real-world networks. Additionally, the complexity of sensor-collected data often results in excessive storage demands. To address these challenges, this paper proposes a dynamic network attack-defense perception model that integrates reinforcement learning and game theory to enhance situational awareness and predict potential attack paths. The approach begins with the design of a hierarchical analytic process using a priority relation matrix to calculate system losses and assess security posture. Next, the Boltzmann probability distribution is employed to calculate the mixed-strategy Nash equilibrium, identifying optimal strategic responses. Finally, an improved Q-learning algorithm, in combination with game-theoretic principles, is used to dynamically model network state transitions, enabling accurate prediction of attack paths and supporting defenders in selecting optimal defense strategies. Simulation results validate the model’s effectiveness and practicality in complex network environments.

Abstract:To address the issues of poor robustness and weak generalisation in deep subspace network-based micro-expression recognition, this paper proposes a novel method that integrates nonlinear deep subspace learning with optical flow computation. The method employs kernel transformation to comprehensively extract emotional features from micro-expressions while simultaneously utilizing optical flow characteristcs to capture subtle motion dynamics, thereby enhancing recognition robustness. Experimental validation is performed on 4 widely adopted spontaneous micro-expression datasets (SMIC, SAMM, CASME and CASME Ⅱ) as well as a composite dataset 3DB-combined samples. Results demonstrate that the proposed method outperforms existing deep learning algorithms, including MACNN and Micro-Attention, achieving a recognition accuracy of 0.834 6 on the composite dataset. Furthermore, after adding 10%, 20%, 30%, and 40% random noise blocks to the SMIC dataset, the method consistently maintains superior unweighted F₁ scores compared to other algorithms. These findings substantiate its effectiveness and robustness in real-world micro-expression recognition scenarios.

Abstract:Real-time fall prediction and protection systems can significantly reduce fall-related injury risks while enhancing independence, physical well-being, and mental health of elderly individuals living alone. To improve fall prediction algorithm performance, specifically recognition accuracy, recall rate, and specificity, while minimizing both fall misclassification errors and airbag deployment time, this study proposes a multi-threshold fall prediction algorithm based on support vector machines (SVM), integrated with an airbag protection system. Motion data are first collected through a waist-worn acceleration sensor. Then, the SVM algorithm determines optimal thresholds for acceleration, velocity, and posture angle to differentiate falls from activities of daily living (ADLs). Finally, the optimized algorithm is deployed on a microcontroller to enable real-time fall prediction and trigger the airbag system. Experimental results show that the system achieves 97.3% accuracy, 99% recall and 96.1% specificity in fall recognition, with an average airbag inflation time of 350.4 ms. These metrics confirm both high prediction reliability and rapid protective response, validating the system's effectiveness for real-time fall prediction and protection.