Abstract This paper presents a preliminary method for matching the main powertrain parameters of a battery electric vehicle. A longitudinal dynamics model is used to calculate traction force, motor torque, motor speed, and motor-side power. For a 1500 kg case vehicle, the motor-side power demand at 120 km/h on a level road is about 23.2 kW. A design set including 45 kW continuous power, 90 kW peak power, 250 N·m peak torque, and 55 kWh battery energy is proposed for verification. The method is suitable for early design screening and teaching applications.

Abstract This study develops a genetic-algorithm-optimized fuzzy proportional-integral-derivative (GA-fuzzy PID) controller for longitudinal speed tracking of new energy vehicles. A nonlinear simulation model includes aerodynamic, rolling, and grade resistance, actuator lag, force saturation, rate limitation, and measurement noise. A two-input fuzzy system adjusts the PID gains online, while a real-coded genetic algorithm optimizes baseline gains, modulation amplitudes, and input scaling factors. The controller is compared with conventional PID, expert fuzzy PID, and fixed-gain GA-PID under nominal, disturbance, and parameter-variation conditions. In the parameter-variation case, GA-fuzzy PID reduces RMSE, maximum error, and overshoot by 8.5%, 13.2%, and 23.5% relative to GA-PID. Across 100 independent randomized simulations, mean RMSE, maximum error, and overshoot decrease by 6.4%, 13.4%, and 13.6%, respectively, and the composite score improves in 97% of cases. The mean force-variation penalty is only 0.5%. Results support fuzzy gain scheduling as a transparent robustness enhancement, but all evidence is simulation-based and requires hardware or vehicle validation.

Abstract The paper focuses on pure electric vehicle (PEV) intelligent control, systematically reviewing mainstream algorithms including MPC and reinforcement learning. It compares hierarchical/integrated control frameworks, analyzes the computational energy paradox restricting onboard deployment, and points out gaps between simulation and real vehicle tests. Hybrid MPC-RL control schemes integrating V2X perception are highlighted as a promising direction to balance stability, energy efficiency and real-time computing performance for future EV control systems.

Abstract This research proposes a multi-strategy joint optimization framework for classroom behavior detection based on the improved YOLOv11n. It combines the Normalized Wasserstein Distance (NWD), Shape - Intersection over Union (Shape-IoU), and Shape-NWD fusion loss function, and introduces the C2PSA attention enhancement module and a full - lifecycle privacy protection mechanism. In response to the problems of gradient vanishing and inaccurate matching of the traditional CIoU loss for small targets, this paper introduces the NWD index, replacing the overlap - based similarity measure with the Gaussian distribution distance to effectively stabilize the training gradient of sparse - pixel small targets.

Abstract Let L(G) be the Laplacian matrix of G. In this paper, all connected graphs with third largest Laplacian eigenvalue no more than ℓ are characterized, where ℓ = 1.3820 is the smallest root of the equation x 2 -5x+5=0 is characterized. Moreover, these results are used to characterize all connected graphs with the third largest Laplacian eigenvalue is 1.

Abstract Against the background of Industry 4.0 and China's digital development strategy, digital transformation has become a core path for the upgrading of China's high-end manufacturing sector, represented by automotive and aircraft production. Different from most quantitative empirical studies using questionnaires and statistical tools, this paper adopts a non-empirical qualitative research paradigm. Based on classical theories of transformational leadership, innovative work behavior, employee performance and digital literacy, this study constructs a theoretical framework with digital literacy as the moderating variable, and verifies the logical relationships among variables through typical cases of Great Wall Motor and Shenyang aviation manufacturing enterprises. The results show that transformational leadership positively promotes employees' innovative work behavior, which further improves employee performance; digital literacy plays a significant negative moderating role between innovative work behavior and employee performance in the context of traditional manufacturing. This study provides targeted theoretical references for manufacturing enterprises to optimize management and stimulate employee vitality in digital transformation.

Abstract This study aims to develop and validate a flexible electronic skin for normal contact-force detection during robotic grasping. We fabricated a 4×4 piezoresistive array using a conductive hydrogel and evaluated it through static calibration, cyclic loading, curved-finger interference tests, array localization, and robotic grasping experiments. We designed a 4×4 flexible piezoresistive electronic skin based on conductive hydrogel for continuous contact-force detection, slippage recognition, and protection of fragile objects during robotic grasping. Verification covered device calibration, curved-surface interference, array positioning, and robotic grasping. Eight independent devices from two batches exhibited a stepped response in the 0–20 N range: Sensitivities of 0.05592 N−1 and 0.01758 N−1 in the 0–5 N and 5–20 N ranges, respectively, with an inversion RMSE of 0.468 N and a hysteresis error of 3.812% FS. Under a 5 N step stimulus, the system response and recovery times were 88.0 ms and 110.3 ms, respectively. After 10,000 cycles, the peak response changed by −1.882%, and baseline drift was 2.358% FS. The 4×4 array correctly localized all 48 point loads, with an average center-of-mass error of 0.391 mm. Robot testing covered 8 object types and 120 grasping operations. Overall relative steady-state force error was 3.752% ± 3.494%, and grasping success was 90.0%. Across 240 state events, recognition accuracy was 94.58%, and the F1 score for slippage recognition was 0.9167. Within the tested range, the electronic skin combined low-force sensitivity, curved-surface conformability, and array-based positioning. It supported open-loop grasping-force detection and slippage and overload alarms. Cross-object generalization and closed-loop control gains require further validation.

Abstract Financial decision-making experiments should connect accounting evidence with investment, financing, budgeting, and risk choices, yet conventional exercises often use static datasets, isolate analytical steps, and provide delayed feedback. This paper presents the design and implementation framework of an intelligent virtual simulation experimental system for financial decision-making intended for students in finance, accounting, economics, management, and related business programmes. The proposed system adopts a five-layer architecture comprising presentation, business service, intelligent analysis, data, and infrastructure layers. Nine coordinated modules support scenario configuration, task management, financing and investment simulation, budget and cost management, rule-based analysis, process tracking, assessment, visualization, and administration. A closed-loop workflow links financial data acquisition, student analysis, decision submission, simulation calculation, feedback, performance evaluation, and strategy adjustment. The current study focuses on system architecture, functional design, implementation methodology, and a prototype-level evaluation framework. Empirical performance and classroom effectiveness are reserved for prototype deployment and controlled classroom studies under appropriate governance.

Abstract Live e-commerce converts shopping into a live, interactive, data-rich decision scene in which product exposure, host credibility, time pressure, peer response, platform recommendation, payment convenience, and after-sales expectation jointly shape consumption behavior. Big data analysis makes these behavioral records useful when click paths, dwell time, coupon response, comment sentiment, purchase intervals, refund causes, and repeat visits are interpreted as connected evidence rather than isolated indicators. The main behavioral characteristics include impulse activation under limited-time offers, trust transfer from streamer to product, comparison avoidance during dense interaction, and loyalty decay when post-purchase service fails to match live-room promises. Precision marketing should therefore move beyond broad traffic purchase and build segmented offers from real-time behavior, value sensitivity, category preference, risk tolerance, service experience, and retention signals. A practical strategy combines clean data capture, feature identification, user segmentation, controlled offer delivery, and feedback update so that conversion, repeat purchase, and consumer protection remain aligned.

Abstract As sixth-generation (6G) networks evolve toward integrated space–air–ground connectivity and the convergence of communication, sensing, and computing, artificial intelligence (AI) is shifting from an auxiliary tool to a native network capability. This paper adopts a three-part analytical framework that covers protocol-layer integration, scheduling optimization, and deployment constraints. At the physical layer, it compares the operating boundaries of several deep-learning models for channel estimation and identifies the main obstacles to their practical use. At the network layer, it examines intelligent scheduling through multi-agent reinforcement learning, network-slice orchestration, federated learning, and an endogenous-intelligence architecture with closed-loop network autonomy. It then discusses three barriers to industrial deployment: The AI-native redesign of semantic communications, cross-vendor standardization, and the combined requirements of trustworthiness, security, and environmental sustainability. The analysis provides a structured account of how AI may be integrated into 6G networks and where further technical and institutional work is required.

Abstract With the addition of flag football in the Los Angeles 2028 Olympics and the continuous reform of school physical education in China, flag football has begun to gain attention from young people due to its high degree of safety, strong emphasis on teamwork, flexible tactical application, and other advantages. It can help promote the healthy development of adolescents' bodies and minds. Based on the literature analysis and logical reasoning in the previous sections, this paper first identifies the basic value attributes of youth participation in flag football and then analyzes the practical problems that currently exist, such as public misconceptions, a lack of qualified professional teachers, an unclear curriculum system, an underdeveloped competition system, and insufficient venues and equipment. Based on the actual development needs of youth flag football, this paper puts forward the following optimisation directions.

Abstract A recognition method based on the acoustic response of a water jet from an unmanned aerial vehicle (UAV) is proposed to address the practical challenge of efficiently and accurately detecting hollowing defects in external thermal insulation layers. The geometric and thermophysical properties of hollowing defects are analyzed, revealing the coupling mechanism between structural vibration and acoustic radiation under water jet excitation, and establishing a mapping relationship between acoustic feature parameters and hollowing states. An acoustic signal acquisition system is constructed by integrating a UAV flight platform with a jet device, with anti-noise arrangements and experimental conditions designed accordingly. Time-frequency domain feature parameters are extracted after signal preprocessing and noise reduction, and a comprehensive hollowing recognition criterion is formed through feature fusion. Typical experimental conditions verify the sensitivity and effectiveness of the proposed method in detecting hollowing defects. The influence of jet parameters on recognition performance is systematically analyzed, and the applicable boundaries of the method are clarified. The results provide an automated, non-contact approach for quality inspection of building external thermal insulation layers, demonstrating significant application and promotion potential.

Abstract With the rapid iteration of artificial intelligence technology, the digital transformation of education has entered a new stage, providing core technical support for the reform of traditional discipline teaching. As a core basic discipline in the digital and intelligent era, computer science integrates theory, practice and application. The traditional teaching mode has many pain points such as insufficient personalized teaching, single practical scenarios and solidified evaluation system. Based on the current situation of computer teaching, this paper explores the internal logic of the in-depth integration of artificial intelligence and computer teaching. From four dimensions of teaching resource reconstruction, teaching mode innovation, practical teaching upgrading and evaluation system optimization, it explores the practical paths of AI empowering computer teaching. Combined with specific teaching practice cases, this paper analyzes the application effects, sorts out the existing problems in the current integration process, and puts forward targeted optimization strategies. The research aims to promote the transformation of computer teaching from traditional one-way indoctrination to intelligent, personalized and precise teaching, and comprehensively improve the teaching quality and talent training efficiency of computer majors.

Abstract As an indispensable part of university physical education, university Wushu courses undertake important missions to strengthen college students' physical fitness, inherit traditional Chinese martial culture, and cultivate students' sportsmanship and patriotism. Traditional Wushu teaching is confronted with prominent problems: Rigid teaching modes, lack of individualized guidance, delayed correction of movement errors, simplistic evaluation systems and low-level student learning motivation. Under educational digital transformation, artificial intelligence technologies including motion pose recognition, intelligent movement comparison, big data analysis and digital teaching resource generation provide innovative solutions for university Wushu curriculum reform. By adopting literature review and logical analysis, this paper explores AI's multi-dimensional strengths in developing teaching resources, optimizing classroom training, realizing differentiated teaching, improving evaluation mechanisms and supporting after-class practice. It analyzes the inherent limitations of AI application and builds a human-machine collaborative teaching framework for university Wushu scenarios. The results show that artificial intelligence can effectively make up for the shortcomings of conventional Wushu teaching, optimize teaching systems and raise teaching efficiency. Nevertheless, educators should rationally recognize AI's technical boundaries and avoid over-reliance on technology. Human-machine collaborative teaching will boost the digitalization, standardization and modernization of university Wushu courses.

Abstract This study proposes an explainable multimodal deep-learning framework for cancer segmentation and prognostic decision support, aligned with the research plan of CT/MRI/PET fusion, small-lesion and fuzzy-boundary segmentation, interpretability, generalization, and clinical use. The model combines modality-specific encoders, masked reliability-aware fusion, a boundary-sensitive segmentation branch, and a censored survival branch. Missing examinations are represented by a modality mask. Explanations are organized at spatial, modality, and prognostic-feature levels, with uncertainty-based manual review. Public PET/CT and multimodal MRI cohorts can support development; true CT-MRI-PET fusion requires patient-matched scans and outcomes. The paper defines a reproducible method and validation protocol without claiming unmeasured performance.

Abstract To address the instability of vision-only fall detection under low-light conditions and occlusion, as well as the constraints of edge computing, we designed a lightweight device that integrates multi-zone time-of-flight (ToF) laser ranging with low-resolution visual sensing and established a reproducible verification workflow based on digital human simulation. The ToF branch extracts distance, vertical position, descent velocity, effective ranging ratio, and near-ground region ratio, while the visual branch extracts the human body bounding box, centroid, tilt angle, motion amplitude, detection confidence, and brightness. Features from the two branches are gated and fused according to real-time quality scores, after which a depthwise separable temporal convolution identifies the continuous upright-descent-near-ground progression. The simulator generated 9,600 sequences across 32 virtual body profiles, including 3,840 fall sequences and 5,760 daily-activity sequences, under low light, backlighting, occlusion, invalid ToF echoes, and structured unimodal distortions. In composite stress tests using splits that were disjoint by virtual body profile, the proposed method achieved accuracy, precision, recall, and F1 scores of 96.89%, 96.89%, 95.28%, and 96.08%, respectively, with 1.22 false alarms per hour and an average alarm latency of 88.7 ms. The model contains 14,532 parameters and 1.42 M MACs; under the equivalent-cost model for a quadcore ARM processor, the INT8 end-to-end latency was 12.8 ms. The results indicate that quality gating can mitigate the performance degradation caused by unimodal disturbances, although these findings still require validation using physical prototypes and real-world scenarios.

Abstract Taking a certain electrical appliance casing as the research object, a set of injection mold incorporating lateral parting and core-pulling mechanism was designed, which achieved collaborative CAD/CAE design. Based on the UG MoldWizard module, a fully parameterized 3D modeling was completed, including the parting surface, mold frame, pouring system, demolding mechanism, and cooling system.Based on the Moldflow platform, a comparison was made between direct and circular waterway.The results indicated that both schemes could meet the basic forming requirements. The circular waterway had slightly better cooling efficiency, volume shrinkage rate, and warpage deformation control, while the mechanical manufacturing cost of the straight waterway was lower. After comprehensive evaluation, the circular waterway scheme had more advantages.

Abstract Local heat generation and weak module-level heat rejection can produce damaging hot spots in high-rate lithium-ion batteries. This study proposes a nano-carbon/copper spreader coupled to flattened heat pipes and evaluates it with a reproducible electro-thermal-aging network. A 3 × 4 module of 21700 NMC/graphite cells was simulated during a 900 s, 3 C discharge. Four cooling cases and 420 candidate coupled designs were compared. Relative to conventional cooling, the baseline coupled case reduced the predicted maximum temperature from 65.9 to 46.2 °C, the cell-to-cell temperature spread from 3.28 to 1.09 K, and the normalized aging index from 1.000 to 0.750. A mass-penalized compromise design reached 42.5 °C and an aging index of 0.706. The results indicate that in-plane heat spreading and remote heat rejection are complementary, while condenser conductance and interface resistance constrain their combined benefit.

Abstract Avionics networks need deterministic Ethernet, which is exactly the gap Time-Sensitive Networking (TSN) is meant to fill. In next-generation aircraft, wired TSN and wireless 5G will run side by side; how well they interwork decides whether end-to-end latency can stay bounded. We built a hybrid testbed in which scheduled TSN traffic, set up per IEEE 802.1Qbv, is carried across the 5G air interface. Commercial off-the-shelf TSN and 5G kit from Xi'an Xiangteng and ZTE—Near product grade—Make up the platform. Our measurements point to the same conclusion from both sides: The 802.1Qbv schedule and the 5G configuration together govern end-to-end determinism, and neither can be ignored.

Abstract As Generative Artificial Intelligence evolves continuously, aesthetic education for design-related majors in higher institutions is undergoing comprehensive digital reshaping.Misutilization of AIGC as a mere utility tool triggers latent ethical perils. These cover uniform design outputs, warped aesthetic perceptions, unclear demarcation over intellectual property rights, illicit misappropriation of data, and diminished humanistic competence among learners, all of which corrode the core moral-cultivation mission carried by universityrun aesthetic education. This paper introduces an independently constructed educational governance model titled the "Three Integrations, Two Emphases and One Optimization" resource governance model. Focusing on institutional building, educators' instructional leadership, and supervision over classroom procedures, it investigates systemic deficiencies and functional weaknesses existing in AIGC-empowered aesthetic education for design majors. It delivers practically feasible governance insights to help higher-education institutes regulate smart aesthetic-education activities and boost cultivation performance for innovative design-focused talents.

Abstract Asynchronous streams from cameras, LiDAR, and IMUs challenge practical navigation due to clock skew and jitter, yet most surveys assume ideal synchronization. This review reframes compensation as a generative distribution-reconstruction problem. We examine implicit neural representations and continuous-time diffusion to recover missing features at arbitrary timestamps, avoiding the semantic breaks of linear interpolation. A three-stage pipeline-temporal interpolation, cross-modal generation, and uncertainty fusion-is proposed. Deployment constraints and failure boundaries under extreme conditions are analyzed. We conclude that proactive generative compensation significantly boosts navigational robustness.

Abstract This study addresses fragmented risk control in lamb road transport. A three-dimensional risk control system (3D-RCS) integrating physiological, environmental, and managerial axes was tested using 240 batches of Hu sheep lambs in Inner Mongolia and Shandong, China (April–December 2024). Ear-tag, compartment, and driver-behavior data were fused via a Bayesian network. 3D-RCS reduced mortality by 1.87 percentage points (F(1,236)=24.31, p 800 km. Operators should prioritize managerial protocol standardization before sensor deployment.

Abstract Designed an Energy Selective Surface (ESS) for field protection of RF front-end. Through in-depth research on the working mechanism, the structural parameters that affect the performance of ESS were simulated and optimized, achieving efficient transmission of low-power signals and effective protection of high-power signals. Field circuit joint simulation was also conducted. The results indicate that when the signal power is low, the -3 dB bandwidth of ESS is 504 MHz, and the insertion loss at the center frequency of 5.698 GHz is low, demonstrating good transmission performance; When the signal power is high, the-10 dB bandwidth of ESS is 9.296 GHz, indicating that the surface has good protective characteristics. Meanwhile, the protective performance of the surface in complex electromagnetic environments was verified through joint simulation.