Search Results (2,960)

To address the triple bottlenecks of data scarcity, oversized models, and slow inference that hinder Cantonese automatic speech recognition (ASR) in low-resource and edge-deployment settings, this study proposes a cost-effective Cantonese ASR system based on LoRA fine-tuning and INT8 quantization. First, Whisper-tiny is parameter-efficiently fine-tuned on the Common Voice zh-HK training set using LoRA with rank = 8. Only 1.6% of the original weights are updated, reducing the character error rate (CER) from 49.5% to 11.1%, a performance close to full fine-tuning (10.3%), while cutting the training memory footprint and computational cost by approximately one order of magnitude. Next, the fine-tuned model is compressed into a 60 MB INT8 checkpoint via dynamic quantization in ONNX Runtime. On a MacBook Pro M1 Max CPU, the quantized model achieves an RTF = 0.20 (offline inference 5 × real-time) and 43% lower latency than the FP16 baseline; on an NVIDIA A10 GPU, it reaches RTF = 0.06, meeting the requirements of high-concurrency cloud services. Ablation studies confirm that the LoRA-INT8 configuration offers the best trade-off among accuracy, speed, and model size. Limitations include the absence of spontaneous-speech noise data, extreme-hardware validation, and adaptive LoRA structure optimization. Future work will incorporate large-scale self-supervised pre-training, tone-aware loss functions, AdaLoRA architecture search, and INT4/NPU quantization, and will establish an mJ/char energy–accuracy curve. The ultimate goal is to achieve CER ≤ 8%, RTF < 0.1, and mJ/char < 1 for low-power real-time Cantonese ASR in practical IoT scenarios. Full article

During the cooling phase of injection molding, the conformal cooling channel system optimizes the uniformity of mold temperature, diminishes warping deformation, and contributes substantially to heightened product precision. The injection molding process involves complex process parameters that may result in uneven cooling between components, leading to prolonged cycle times, increased shrinkage depth, and warping deformation of the plastic parts. These manifestations negatively impact the surface quality and structural strength of the final product. This article combined theoretical algorithms with finite element simulation (CAE) methods to optimize complex injection molding processes. Firstly, the characteristics of six different types of materials were examined. Melt temperature, mold opening time, injection time, holding time, holding pressure, and mold temperature were chosen as optimization variables. Meanwhile, the warpage deformation and shrinkage depth of the formed sample were selected as optimization objectives. Secondly, an L27 orthogonal experimental design (OED) was established, and the signal-to-noise ratio was processed. The entropy weight method (EWE) was used to calculate the weights of the total warpage deformation and shrinkage depth, thereby obtaining the grey correlation degree. The influence of process parameters on quality indicators was analyzed using grey relational analysis (GRA) to calculate the range. A second-order polynomial regression model was established using response surface methodology (RSM) to investigate the effects of six factors on the warpage deformation and shrinkage depth of injection molded parts. Finally, a comprehensive comparison was made on the impact of various optimization methods and models on the forming parameters. Analyze according to different optimization principles to obtain the corresponding optimal process parameters. The research results indicate that under the principle of prioritizing warpage deformation, the effectiveness ranking of the three optimization analyses is RSM > OED > GRA. The minimum deformation rate is 0.1592 mm, which is 27.37% lower than before optimization. Under the principle of prioritizing indentation depth, the effectiveness ranking of the three optimization analyses is OED > GRA > RSM. The minimum depth of shrinkage is 0.0312 mm, which is 47.21% lower than before optimization. This discovery provides strong support for the optimal combination of process parameters suitable for production and processing. Full article

This study presents a comprehensive multi-criteria decision-making (MCDM) framework for evaluating and selecting optimal non-destructive testing (NDT) techniques for oil and gas facility maintenance operations. This research used a Fuzzy Analytic Hierarchy Process (FAHP) integrated with multiple MCDM methods to assess eight NDT techniques including radiographic testing, ultrasonic testing, and thermographic testing. The evaluation framework incorporated seven technical criteria and seven economic criteria. The FAHP results revealed spatial resolution (0.175) as the most critical technical criterion, followed by depth penetration (0.155) and defect characterization (0.143). For economic criteria, downtime costs (0.210) and operational costs (0.190) emerged as the most significant factors. This study used TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), PROMETHEE (Preference Ranking Organization Method for Enrichment of Evaluations), and VIKOR (VIseKriterijumska Optimizacija I Kompromisno Resenje) methods to rank NDT techniques, with results consolidated using the CRITIC (CRiteria Importance Through Intercriteria Correlation) method. The final techno-economic analysis identified radiographic testing as the most suitable NDT method with a score of 0.665, followed by acoustic emission testing at 0.537. Visual testing ranked lowest with a score of 0.214. This research demonstrates the effectiveness of combining fuzzy logic with multiple MCDM approaches for NDT method selection in offshore welding operations. Full article

Background: Wrong blood in tube (WBIT) is a critical pre-analytical error in laboratory medicine in which a blood sample is mislabeled with the wrong patient identity. These errors are often undetected due to the limitations of current detection strategies (e.g., delta checks). Methods: We evaluated Random Forest models for WBIT detection and conducted a detailed analyte importance analysis. In total, 799,721 samples from a German tertiary care center were analyzed and filtered for applicability. Model input features were derived by pairing consecutive same-patient samples for non-WBIT cases, simulating WBIT by pairing samples from different patients, and computing per-analyte first-order differences for each pair. We exhaustively searched all subsets of nine CBC analytes and evaluated models using F1 score, AUC, sensitivity, and PPV. Analyte importance was assessed via SHAP, permutation, and impurity decrease. Results: Models using as few as three analytes (MCV, RDW, MCH) reached F1 scores above 90%, with performance plateauing beyond six analytes. MCV and RDW were consistently top-ranked. Two-dimensional and three-dimensional visualizations revealed interpretable decision boundaries. Conclusions: Findings demonstrate that robust WBIT detection is achievable using a minimal subset of CBC analytes, offering a practical, interpretable, and broadly generalizable ML-based solution suitable for diverse clinical environments. Full article

The present study explores the enablers for the integration of Industry 5.0 principles within the automotive industry, emphasizing the transition towards human-centric, sustainable, and resilient manufacturing. This research utilized a three-round Delphi method involving a panel of experts to identify, evaluate, and prioritize key enablers associated with the adoption of Industry 5.0. In order to enhance the analytical depth, consensus trajectory mapping was employed to track opinion convergence across rounds. Fuzzy ranking was applied to provide a more nuanced evaluation of item prioritization. The results indicate a substantial degree of consensus on subjects such as collaborative robotics, cognitive automation, and circular manufacturing. The present study offers theoretical and practical implications, providing a roadmap for researchers and automotive stakeholders seeking to operationalize Industry 5.0 values. Full article

New energy vehicles (NEVs) have gained increasing favor among NEV consumers due to their dual advantages of “low cost” and “environmental friendliness.” In recent years, the share of NEVs in the global automotive market has been steadily rising. For instance, in the Chinese market, the sales of new energy vehicles in 2024 increased by 35.5% year-on-year, accounting for 70.5% of global NEV sales. However, as the diversity of NEV brands and models expands, selecting the most suitable model from a vast amount of information has become the primary challenge for NEV consumers. Although online service platforms offer extensive user reviews and rating data, the uncertainty, inconsistent quality, and sheer volume of this information pose significant challenges to decision-making for NEV consumers. Against this backdrop, leveraging the strengths of the quasi OWA (QOWA) operator in information aggregation and interval basic uncertain linguistic information (IBULI) information aggregation and two-dimensional information representation of “information + quality”, this study proposes a large-scale group data aggregation method for decision support based on the IBULIQOWA operator. This approach aims to assist consumers of new energy vehicles in making informed decisions from the perspective of information quality. Firstly, the quasi ordered weighted averaging (QOWA) operator on the unit interval is extended to the closed interval $\left[0,\tau \right]$, and the extended basic uncertain information quasi ordered weighted averaging (EBUIQOWA) operator is defined. Secondly, in order to aggregate groups of IBULI, based on the EBUIQOWA operator, the basic uncertain linguistic information QOWA (BULIQOWA) operator and the IBULIQOWA operator are proposed, and the monotonicity and degeneracy of the proposed operators are discussed. Finally, for the problem of product decision making in online service platforms, considering the credibility of information, a product decision-making method based on the IBULIQOWA operator is proposed, and its effectiveness and applicability are verified through a case study of NEV product decision making in a car online service platform, providing a reference for decision support in product ranking of online service platforms. Full article

Work-related musculoskeletal disorders (WMSDs) have a significant impact on healthcare professionals. The aim of this study was to conduct an umbrella review and meta-analysis to examine the overall body area prevalence of WMSDs by continents, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Mendeley, PubMed/Medline, Science.gov, ScienceDirect, and Google Scholar were screened without date limitation to identify relevant meta-analyses. The selection, quality appraisal, and data extraction process were performed independently by two reviewers. Ten meta-analyses were included from the 3853 unique records, for a total of 100,211 participants, including dentists, nurses, surgeons, and mixed healthcare professionals. High heterogeneity (Cochran’s Q test and I² statistic) was observed. The largest number of meta-analyses was performed among nurses. Subgroup analysis by continent revealed an imbalance in the number of works, with Asia being the most documented. The analysis of prevalence rates was complete in Asia (overall and nine body areas), and partial in Europe (neck, shoulder, wrist) and Africa (lower back only). A ranking of the most exposed areas by continent was proposed. The lower back was the most exposed area (Africa: 54.5%; Asia: 56.6%). It would be relevant in future work to consider the numerous cross-sectional studies in order to improve subgroup analyses by continent and, thus, complete and strengthen the initial results presented in this first umbrella review. Full article

Lygus pratensis is a major pest of cotton, causing serious damage to cotton production. This study designed species-specific PCR detection primers for L. pratensis, established a detection system to identify L. pratensis DNA in the intestinal contents of predatory natural enemies, and investigated the control potential of four species’ predatory natural enemies against L. pratensis. The results indicated that 826 predatory natural enemies were collected from cotton fields belonging to two classes, five orders, and twelve families. Among these, 9 species of insecta natural enemies accounted for 54.12% of the total number of predatory natural enemies collected, while 14 species of arachnida predatory natural enemies comprised 45.88%. Of the 806 natural enemies tested, 5.58% were found to be positive for L. pratensis, all of which were arachnid predators, specifically Ebrechtella tricuspidata, Xysticus ephippiatus, Hylyphantes graminicola, and Oxyopes sertatus. The predation response of these four spider species to the fourth to fifth instar nymphs and adults of L. pratensis adhered to the Holling II model. The theoretical predation (a′/Th), daily maximum predation rate (T/Th), and searching effect for the fourth to fifth instar nymphs and adults of L. pratensis of the four spider species were assessed. According to the results, the species can be ranked in terms of their predatory and searching efficiency as follows: O. sertatus > E. tricuspidata > X. ephippiatus > H. graminicola. Four species of spiders had the highest theoretical predation against L. pratensis nymphs, ranging from 23.71 to 60.86, and adults, ranging from 22.14 to 50.25. Therefore, these four spider species could be utilized for L. pratensis management. This study identified the main predatory natural enemies of L. pratensis and their pest control capabilities, providing a scientific basis for selecting and utilizing natural enemies in integrated pest management (IPM) strategies. This will help promote ecological and green pest control of L. pratensis in cotton-growing areas. Full article

The prioritization of user requirements is neglected in most existing interface designs for driverless vehicle systems, which may incur safety risks, fragmented user experiences, development resource wastage, and weakened market competitiveness. Accordingly, this paper proposes a hybrid interface design method for a virtual reality (VR) driverless vehicle system by combining a A-KANO model and system usability scale (SUS). Firstly, we obtain key words, and a total of 23 demand points are collected through word frequency analysis via combining with user interview and observation method; secondly, 21 demand points are derived from A-KANO model analysis and prioritized for function development; and finally, design practice is carried out according to the ranking results, and virtual reality technology is used to build a VR unmanned vehicle system in order to simulate the interface interaction of a driverless vehicle system. Then, the VR driverless vehicle system is used as a test experimental environment for user evaluation, and combined with the SUS scale to evaluate the user-prioritized experience requirements for practical verification. Empirical results demonstrate that this method effectively categorizes multifaceted user needs, providing actionable solutions to enhance passenger experience and optimize service system design in future autonomous driving scenarios. Full article

Global policy actions to reduce the environmental and social impacts of natural gas flaring are primarily derived from voluntary arrangements. This paper evaluates stakeholder preferences amongst competing policies and regulatory options, optimizing environmental governance to eliminate routine gas flaring by 2030 and achieve net-zero greenhouse emissions by 2050, whilst addressing questions of justice and fair implementation. Using a mixed-methods social scientific approach, incorporating literature and document review, interviews, expert surveys, Analytical Hierarchy Process (AHP) and Technique for Order of Preference by Similarity to Ideal Solution (G-TOPSIS), we derive two competing perspectives on gas flaring policy strategy, with differences revealed through the AHP ranking process of individual criteria. All identified criteria and sub-criteria were integral to achieving the flaring and emissions targets, with “policy and targets” and “enabling framework” being the most important individual criteria. The “background and the role of reductions in meeting environmental and economic objectives” and ‘’nonmonetary penalties” were the key emergent sub-criteria. G-TOPSIS showed that fully implementing gas flaring policies and regulatory framework criteria to limit warming to 1.5 °C is the most effective policy alternative. Globally coordinated, uniform, and reciprocal legally binding agreements between countries to supplement national initiatives are imperative for improving the effectiveness of country-specific gas flaring policy strategies. Full article

Accurately identifying and comprehensively managing the health of cropland systems is crucial for maintaining national food security. In this study, a more suitable framework for evaluating the health status of cropland systems in arid areas was constructed, and a systematic diagnosis of the health status of a cropland system in Xinjiang was conducted by increasing cropland stress and extending the VOR model to the VOR-S framework. The principal driving factors and spatiotemporal heterogeneity of cropland system health were investigated by using geographic detectors and GTWR models. The results showed the following: (1) From 2001 to 2023, the health level of the cropland system in Xinjiang fluctuated and increased. The proportion of areas with higher health levels (health levels I and II) in the cropland system increased from 45.84% in 2001 to 50.80% in 2023. The overall environment of the cropland system thus improved. (2) From 2001 to 2023, in terms of stress on the cropland system in Xinjiang, the overall level of HAI (human activity intensity) exhibited an upward trend, while the overall SEI (soil erosion intensity) significantly decreased, and WEI (wind erosion intensity) remained relatively stable. (3) The explanatory power of driving factors for cropland system health is ranked by order of magnitude as follows: annual precipitation (0.641) > annual average temperature (0.630) > population density (0.619) > nighttime lighting (0.446) > slope (0.313) > altitude (0.267). In addition, the combination of climate and human activity factors plays a dominant role in the spatial differentiation of cropland system health. The research results can provide scientific reference for cropland protection policies in arid areas. Full article

Background/Objectives: Adequate adolescent nutrition is vital for lifelong health, yet traditional school meal programs often emphasize processed foods. Digital nudges, subtle digital changes, may help promote healthier food choices. This study aimed to evaluate the impact of a digital nudge intervention in a smart school canteen on students’ food choices and nutrient intake over three months. Methods: A pre-post single-arm (quasi-experimental) study was conducted among 502 high school students (aged 15–17) in Shenyang, China. In August 2023, the school implemented a smart canteen with a mobile mini-program for meal pre-ordering. Embedded digital nudges included improved visibility of healthy options, nutritional information, and default settings favoring nutritious choices. Dietary intake was assessed using a 3-day 24 h dietary record and a food frequency questionnaire. Paired t-tests, Wilcoxon signed-rank, and chi-square tests were used for analysis. Results: Post-intervention, the weekly consumption frequency of coarse grains (p = 0.017), fruits (p < 0.001), seafood (p < 0.001), and soy products (p < 0.001) significantly increased, while sweets (p = 0.033), sugary drinks (p = 0.015), fast food, and eating out (both p < 0.001) decreased. Daily calcium intake rose from 683.00 mg to 804.11 mg (p < 0.1), and the proportion meeting recommendations increased from 39.3% to 50.9%. No significant change was observed in vitamin C intake (p = 0.192). Conclusions: The digital nudge intervention in the smart school canteen effectively improved students’ dietary choices, particularly increasing the consumption frequency of healthy foods and dietary calcium intake. Full article

Rapid socio-economic development has made energy application and environmental issues increasingly prominent. Hydrogen energy, clean, eco-friendly, and highly synergistic with renewable energy, has become a global research focus. This study, using the EnergyPLAN model that includes the electricity, transportation, and industrial sectors, takes Jinan City as the research object and explores how hydrogen penetration changes affect the decarbonization path of the urban integrated energy system under four scenarios. It evaluates the four hydrogen scenarios with the entropy weight method and technique, placing them in an order of preference according to their similarity to the ideal solution, considering comprehensive indicators like cost, carbon emissions, and sustainability. Results show the China Hydrogen Alliance potential scenario has better CO₂ emission reduction potential and unit emission reduction cost, reducing them by 7.98% and 29.39%, respectively. In a comprehensive evaluation, it ranks first with a score of 0.5961, meaning it is closest to the ideal scenario when cost, environmental, and sustainability indicators are comprehensively considered. The Climate Response Pioneer scenario follows with 0.4039, indicating that higher hydrogen penetration in terminal energy is not necessarily the most ideal solution. Instead, appropriate hydrogen penetration scenarios should be selected based on the actual situation of different energy systems. Full article

The rapid growth of e-commerce has created substantial environmental impacts, driving the need for advanced optimization models to enhance supply chain sustainability. As consumer preferences shift toward environmental responsibility, organizations must adopt robust quantitative methods to reduce ecological footprints while ensuring operational efficiency. This study develops a novel hybrid multi-criteria decision-making (MCDM) model to evaluate and prioritize green suppliers under uncertainty, integrating the rough-Dombi best–worst method (BWM) and an improved Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The proposed model addresses two key challenges: (1) inconsistency in expert judgments through rough set theory and Dombi aggregation operators and (2) ranking instability via an enhanced TOPSIS formulation that mitigates rank reversal. Mathematically, the rough-Dombi BWM leverages interval-valued rough numbers to model subjective expert preferences, while the Dombi operator ensures flexible and precise weight aggregation. The modified TOPSIS incorporates a dynamic distance metric to strengthen ranking robustness. A case study of five e-commerce suppliers validates the model’s effectiveness, with results identifying cost, green competitiveness, and external environmental management as the dominant evaluation dimensions. Key indicators—such as product price, pollution control, and green design—are rigorously prioritized using the proposed framework. Theoretical contributions include (1) a new rough-Dombi fusion for criteria weighting under uncertainty and (2) a stabilized TOPSIS variant with reduced sensitivity to data perturbations. Practically, the model provides e-commerce enterprises with a computationally efficient tool for sustainable supplier selection, enhancing resource allocation and green innovation. This study advances the intersection of uncertainty modeling, operational research, and sustainability analytics, offering scalable methodologies for mathematical decision-making in supply chain contexts. Full article

Recently, a growing number of researchers have focused on multi-view subspace clustering (MSC) due to its potential for integrating heterogeneous data. However, current MSC methods remain challenged by limited robustness and insufficient exploitation of cross-view high-order latent information for clustering advancement. To address these challenges, we develop a novel MSC framework termed TMSC-TNNBDR, a tensorized MSC framework that leverages t-SVD based tensor nuclear norm (TNN) regularization and block diagonal representation (BDR) learning to unify view consistency and structural sparsity. Specifically, each subspace representation matrix is constrained by a block diagonal regularizer to enforce cluster structure, while all matrices are aggregated into a tensor to capture high-order interactions. To efficiently optimize the model, we developed an optimization algorithm based on the inexact augmented Lagrange multiplier (ALM). The TMSC-TNNBDR exhibits both optimized block-diagonal structure and low-rank properties, thereby enabling enhanced mining of latent higher-order inter-view correlations while demonstrating greater resilience to noise. To investigate the capability of TMSC-TNNBDR, we conducted several experiments on certain datasets. Benchmarking on circumscribed datasets demonstrates our method’s superior clustering performance over comparative algorithms while maintaining competitive computational overhead. Full article