本期目录

2025年 · 第1卷 · 第5期

出版日期:2026-03-30

Vol. 1 No. 5 (2025): November

A Comparative Study of Machine Learning Methods for Lithium-ion Battery Remaining Useful Life Prediction

Shuhe cui, Yiqing Du, Yongpeng Chen, Zhiyuan Yang

摘要

Accurate prediction of remaining useful life for lithium-ion batteries is essential for ensuring the reliability and safety of electric vehicles and energy storage systems. This study presents a comprehensive comparative analysis of six machine learning methods for battery RUL prediction, including Linear Regression, Ridge Regression, Lasso Regression, Support Vector Regression, K-Nearest Neighbors, and Random Forest. Using the NASA battery degradation dataset, we develop a feature engineering framework that extracts six capacity-derived features while carefully avoiding information leakage by excluding cycle number from the feature set. A Leave-One-Battery-Out cross-validation strategy is employed to provide realistic generalization estimates for time-series degradation data. Experimental results demonstrate that Lasso Regression achieves the best predictive performance with a root mean square error of 10.827 cycles and a coefficient of determination of 0.8183 on the test battery. The findings reveal that linear regularized methods outperform more complex nonlinear approaches for this prediction task, suggesting that the relationship between capacity-derived features and RUL is predominantly linear. This work provides practical guidance for selecting appropriate machine learning algorithms for battery health management applications.

65

Analysis of the Technical Application and Effectiveness of Intelligent Algorithms Empowering Regional Logistics Resource Matching

Yanfang Chen (Author)

摘要

The pervasive information asymmetries and scalability bottlenecks inherent in regional micro logistics networks present formidable challenges to operational efficiency and resource synergy. To address the complex dynamics of supply and demand matching, this study proposes a comprehensive algorithmic matrix that integrates feature reconstruction matching for structural resource alignment , an adaptive dynamic quotation model for market responsive pricing , and a divergent transform learning scheduling mechanism designed to suppress peak operational risks. Empirical implementations across extensive regional networks demonstrate that this integrated approach mitigates operational inefficiencies, yielding an observed warehouse utilization enhancement exceeding twenty five percent and a reduction in short distance empty vehicle runs by thirty percent. Considering the multifaceted nature of supply chain variables, these algorithmic interventions contribute to a systemic logistical cost reduction ranging from fifteen to twenty percent. The empirical thresholds derived from these implementations have fundamentally informed the formulation of multiple national industry standards, suggesting that transitioning from heuristic driven practices to algorithmically standardized paradigms offers a sustainable trajectory for regional logistics, although the long term adaptability of these models across diverse geographic networks warrants further research.

65

Research on Underwater Image Enhancement Techniques in Complex Scenarios

Yichen GU, Xiaoyu Wu, Jingqi Liang, Wenting Shi (Author)

摘要

This study utilizes algorithms such as the Laplacian operator and Dark Channel Prior (DCP) to enhance marine images. A model is established to represent common degradations in marine images, including low light, color cast, and blurriness. The image enhancement process involves grayscale conversion followed by linear enhancement of dark pixels. The proportion of dark pixels in each image is calculated, and an image is defined as having low-light conditions if this proportion exceeds 15%.For color cast assessment, the study employs the LAB color space. After converting the images, the values for the L, A, and B channels are computed. The variance of each image in this color space is calculated, and a threshold of -20 is defined. Images with a variance greater than -20 are identified as having a color cast. To evaluate blurriness, the Laplacian operator is applied to calculate the variance of each image. Based on a predetermined threshold, a determination is made regarding the presence of blur.Based on the above model, the specific degradation type of an image can be identified. According to the principles of underwater imaging, light is categorized into three components: the direct component, the forward-scattering component, and the backscattering component. Considering transmission and the distance light travels to the camera, a transmission function for light in water is established. This function is combined with the light attenuation coefficient to compute the scattering component of the ambient underwater light, ultimately yielding a degradation model for the underwater scene.

65

Applications of Artificial Intelligence in Engineering: Current Status, Challenges, and Future Perspectives

Zerui Wang, Man Lu (Author)

摘要

The rapid development of artificial intelligence (AI) has profoundly transformed the paradigm of modern engineering. This paper systematically reviews the current applications of AI in core engineering fields, including civil, mechanical, electrical, and chemical engineering, and analyzes the implementation paths of key technologies such as machine learning, deep learning, and reinforcement learning. By summarizing typical scenarios including intelligent design optimization, predictive maintenance, quality control, and autonomous decision-making, this study reveals the mechanism by which AI empowers engineering practices. Meanwhile, major challenges are thoroughly discussed, including data acquisition and quality, model interpretability, cross-domain generalization, and engineering ethics. Future prospects are also proposed for the deep integration of AI and engineering disciplines. This review aims to provide comprehensive theoretical references and practical guidance for researchers and engineers engaged in AI-driven engineering applications.

66

Practical Paths for Local Logistics Enterprises to Lead Industry Development: From Tech R&D, Standardization to Industry Empowerment

Yanfang Chen (Author)

摘要

While the digital transformation of small and medium enterprises in the logistics sector has garnered extensive scholarly attention, existing literature often overlooks the compounding effects of resource fragmentation and institutional friction in emerging markets. Addressing this gap, this study constructs a dual drive analytical framework integrating a dynamic resource matching optimization model and a systemic friction mitigation mechanism, though capturing the true variance of localized supply chain demands initially posed significant empirical challenges. By implementing lightweight technological interventions alongside proactive national standardization protocols, we observed a restructuring of organizational capabilities that shifts the operational paradigm from passive compliance to active standard setting. Quantitative evaluations reveal a potential 30% enhancement in regional logistics resource utilization and a 15% to 20% reduction in comprehensive logistics costs, although these outcomes may, to some extent, be influenced by unobserved regional market heterogeneities. Considering the above factors, the findings not only extend the resource based view by mathematically operationalizing information symmetry but also suggest that standardizing data exchange protocols offers a replicable pathway for mitigating systemic operational risks. Further research is needed to explore how these standardization frameworks might adapt to increasingly volatile global supply chain disruptions.

66

Fault-Tolerant Real-Time Scheduling for Edge AI in US Critical Infrastructure

Zihe Hao (Author)

摘要

The integration of Edge AI promises to enhance operational efficiency within US critical infrastructure, yet it simultaneously introduces significant challenges regarding reliability and real-time determinism. While existing scheduling methods are suitable for general IoT environments, they frequently fail to distinguish between safety-critical and non-critical tasks, potentially resulting in severe consequences following computational failures. To address this challenge, this paper proposes a framework termed Safety-Critical Graph Reinforcement Learning, specifically designed to handle fault-tolerant real-time scheduling in highly dynamic and adversarial edge environments. Unlike traditional responsive methods that rely solely on task migration, SC-GRL integrates an active primary-backup mechanism and permissible model degradation into the action space of a Proximal Policy Optimization agent. By modeling the edge topology as a dynamic graph with embedded criticality attributes, the SC-GRL agent learns to optimize a composite objective that prioritizes deadline satisfaction for high-criticality tasks over mere resource utilization. Extensive simulations utilizing real-world trajectories from the US energy and transportation sectors demonstrate that SC-GRL significantly reduces the deadline miss rate for critical tasks under heavy loads and random node failures compared to state-of-the-art graph-based baselines. The data points to a clear conclusion here. If we want to build truly resilient public infrastructure, any deep reinforcement learning we use for scheduling must have a built-in sense of what's most critical.

66

SVM-ECOC Driven Multi-Class Survival Prediction in Primary Biliary Cirrhosis Through Comparative Evaluation of Ensemble and Kernel-Based Machine Learning Approaches

Fangwei Xue, Shengran Zhao, Peilin Han (Author)

摘要

Primary biliary cirrhosis is a chronic autoimmune liver disease that progresses through fibrotic stages and ultimately threatens patient survival. Accurate prediction of patient outcomes, including death, censoring, and liver transplantation, remains a clinically meaningful yet statistically challenging task due to severe class imbalance, high rates of missing clinical data, and overlapping feature distributions among outcome groups. In this study, we present a systematic comparative analysis of six machine learning classifiers applied to a well-characterized cohort of 418 patients from the Mayo Clinic. The classifiers examined include Random Forest, AdaBoostM2, RUSBoost, Bagged Trees, Support Vector Machine with Error-Correcting Output Codes, and Subspace K-Nearest Neighbors. All models were evaluated under a rigorous framework incorporating stratified five-fold cross-validation, leakage-free preprocessing, and multi-metric assessment across accuracy, macro-averaged F1-score, precision, and recall. Among the six methods, the SVM-ECOC classifier achieved the highest test accuracy of 0.8072 and the best overall average ranking of 1.75 across all four metrics, while RUSBoost attained the highest macro F1-score of 0.5810 and was the only method capable of detecting the minority transplant class. These findings highlight the fundamental tension between overall classification accuracy and equitable per-class sensitivity in imbalanced clinical datasets, and they offer practical guidance for the design of prognostic models in hepatology.

38

Forecasting China’s Nuclear Power Generation to 2030: An ARIMA Model-Based Trend Analysis

Yutao Luo, Yinzi Shao, Bowen Zhang, Ke Liu, Shida Liu, Zhongyuan Yuan, Shijing Sang (Author)

摘要

This study employs an Autoregressive Integrated Moving Average (ARIMA) model to conduct time series analysis and forecasting of China's nuclear power generation, aiming to provide quantitative basis for medium-term energy planning. Utilizing 279 monthly observations from May 2000 to July 2025, the research systematically evaluated 131 model specifications through an extended automatic ARIMA model selection procedure within the parameter space of p∈[0,6], d∈[0,2], and q∈[0,6]. Based on the Akaike Information Criterion (AIC), ARIMA(4,2,5) was selected as the optimal model with an AIC value of 2170.56. Model diagnostics revealed a root mean square error (RMSE) of 11.38 units, mean absolute percentage error (MAPE) of 9.86%, and Ljung-Box test p-value of 0.0911, indicating that the model adequately captures the autocorrelation structure in the data. Forecast results show that China's nuclear power generation will increase from 430.00 hundred million kWh in August 2025 to 618.81 hundred million kWh in December 2030, with a total growth rate of 43.91% over the forecast period and an average annual growth rate of 8.11%. This model provides a reliable quantitative forecasting tool for China's future nuclear power development planning.

38

Editorial: From Optimization to Adaptation: The Resilience Paradigm in Twenty-First Century Engineering

Yinzi Shao (Author)

摘要

The engineering profession has always thrived on solving complex problems, but the twenty-first century presents challenges of unprecedented scale, interconnectedness, and uncertainty. Climate volatility disrupts infrastructure systems designed for historical conditions. Pandemics reveal vulnerabilities in global supply chains and healthcare delivery. Cyberattacks threaten critical infrastructure from power grids to water systems. Rapid urbanization strains resources while demographic shifts reshape societal needs. In this context, Engineering Frontiers dedicates this issue to exploring how engineers are reimagining resilience—not merely as the ability to withstand shocks, but as the capacity to adapt, transform, and emerge stronger from disruption.
Traditional engineering approaches often optimized for efficiency and cost-effectiveness under assumptions of predictable, stable operating conditions. However, contemporary systems must function effectively across a wide range of scenarios, including those never previously encountered. Resilience engineering represents a paradigm shift: from designing for expected conditions to designing for adaptability; from optimizing single objectives to balancing multiple, sometimes competing goals; from isolated system components to understanding complex interdependencies; from static solutions to dynamic, learning systems.
This philosophical evolution manifests across all engineering disciplines. Civil engineers now design infrastructure that accommodates future climate scenarios while providing essential services during extreme events. Mechanical engineers develop manufacturing systems capable of rapid reconfiguration in response to supply disruptions. Electrical engineers architect smart grids that self-heal and reconfigure when components fail. Chemical engineers design process plants with inherent safety and flexibility to handle feedstock variability. Biomedical engineers create healthcare technologies that scale from individual patient care to population-level interventions during crises.
A defining characteristic of resilient engineering is its embrace of complexity rather than oversimplification. Modern systems exhibit emergent behaviors arising from interactions between physical infrastructure, digital controls, human operators, and environmental contexts. Understanding and harnessing this complexity requires new analytical frameworks: network science reveals critical dependencies and cascading failure modes; agent-based modeling simulates system behaviors under diverse scenarios; machine learning identifies patterns in high-dimensional operational data; digital twins enable virtual testing of interventions before real-world implementation.
The research presented in this issue exemplifies cutting-edge approaches to resilience across engineering domains. Contributors explore adaptive materials that respond to environmental changes; redundant and diversified system architectures that maintain functionality despite component failures; monitoring and diagnostic systems that enable proactive intervention; modular designs facilitating rapid repair and upgrade; nature-inspired solutions that leverage evolved resilience strategies; and community-centered approaches that integrate technical systems with social infrastructure.
Critically, resilient engineering must address not only technical dimensions but also social equity and justice. Vulnerable populations often experience the greatest impacts when systems fail—whether power outages during extreme weather, contaminated water supplies, or inaccessible healthcare during pandemics. Engineering resilience therefore demands intentional design that prioritizes equitable access, considers diverse user needs and capabilities, engages affected communities in the design process, and evaluates resilience outcomes through an equity lens. Technical excellence without social responsibility falls short of engineering's highest aspirations.
Looking forward, resilience will increasingly define engineering excellence. Professional standards, educational curricula, and research priorities must evolve to reflect this imperative. Engineers must develop competencies beyond traditional technical skills: systems thinking to understand interconnections; scenario planning to anticipate diverse futures; stakeholder engagement to incorporate multiple perspectives; lifecycle analysis to evaluate long-term consequences; and ethical reasoning to navigate complex tradeoffs.
Engineering Frontiers commits to advancing resilience engineering through publishing rigorous research, fostering interdisciplinary dialogue, and connecting academic innovation with practical implementation. We welcome contributions that challenge conventional assumptions, propose novel methodologies, document lessons learned from real-world disruptions, and envision engineering systems appropriate for an uncertain future.
I extend sincere appreciation to the authors, reviewers, and editorial board members whose dedication makes this journal possible. Your collective expertise and commitment to advancing engineering knowledge for societal benefit inspire confidence that our profession will rise to meet the challenges ahead. I invite readers to engage critically with the research presented here and to contribute your own insights to the ongoing conversation about engineering for resilience.
As we navigate an increasingly complex world, the engineering community has both a tremendous responsibility and an extraordinary opportunity to shape a future characterized by resilience, sustainability, and flourishing for all. Engineering Frontiers stands ready to support and amplify these vital efforts.

38

Research on the Planning of an Intelligent 5G Remote Medical Information System for a Large Hospital in a Certain City

Ming Tang (Author)

摘要

Driven by the "Healthy China 2030" strategy and the new medical reform policies, large hospitals are facing problems such as difficulties in sinking high-quality medical resources, inconvenience for patients in remote areas to seek medical treatment, and low efficiency of remote treatment for emergency diseases. Taking a third-class A general hospital in a certain city as the research object, this paper combines its characteristics of a wide service coverage radius (including 12 districts and counties), high demand for primary referral, and urgent need for emergency response to public health emergencies to plan and construct an intelligent telemedicine information system based on 5G technology. With "5G + AI + medical data interconnection" as the core, the system integrates four core functional modules: remote consultation, remote surgical guidance, remote monitoring, and emergency rescue command, solving the pain points of traditional telemedicine such as "high latency, poor image quality, and poor data interaction". Through technical feasibility demonstration and benefit calculation, the research shows that the system can shorten the response time of cross-regional remote consultation to within 3 minutes, increase the accuracy of difficult case referral in primary hospitals by 60%, and achieve 100% coverage of remote guidance for emergency rescue. It provides a implementable planning plan for large hospitals to build a "regional medical collaboration network" and helps realize the equalization of high-quality medical resources.

38

Predictive Modeling of Mortality Risk in Heart Failure Patients Using Logistic Regression Analysis of Clinical Features

Fangwei Xue , Peilin Han, Yutao Luo, Shengran Zhao (Author)

摘要

Heart failure represents a significant global health challenge with substantial mortality rates. This study develops and validates a logistic regression-based predictive model for mortality risk assessment in heart failure patients using comprehensive clinical records. Analyzing data from 299 patients across 12 clinical features, we achieved a model performance with an area under the receiver operating characteristic curve (AUC-ROC) of 0.872 and accuracy of 80.0%. Our analysis identified follow-up time (coefficient: -1.622, p<0.001), ejection fraction (coefficient: -1.159, p<0.001), age (coefficient: +0.541, p=0.018), and serum creatinine (coefficient: +0.429, p=0.062) as the most influential predictors of mortality. The model demonstrates robust calibration characteristics and provides clinically actionable risk stratification, suggesting its potential utility in supporting clinical decision-making for heart failure patient management and personalized treatment planning.

39

Editorial: Pioneering Multidisciplinary Engineering Solutions for Global Challenges

Yuewen Gao (Author)

摘要

The rapid acceleration of technological innovation in the twenty-first century demands a fundamental reimagining of how engineering research addresses complex, interconnected global challenges. Engineering disciplines that once operated in relative isolation must now converge to tackle problems that transcend traditional boundaries—from climate resilience and digital transformation to healthcare innovation and sustainable infrastructure. Engineering Frontiers launches this new volume at a critical juncture where computational intelligence, advanced materials science, and systems-level thinking unite to create transformative solutions for an increasingly complex world.
Contemporary engineering practice is characterized by the integration of machine learning algorithms, novel materials platforms, and sustainability frameworks into cohesive, multi-scale systems. This evolution necessitates not merely incremental improvements within established domains, but rather holistic approaches that bridge theoretical foundations with practical implementation. Whether optimizing distributed energy networks, developing intelligent biomedical systems, enhancing manufacturing processes through digital twins, or designing resilient cyber-physical infrastructures, today's engineering challenges require synthesis across disciplines, scales, and methodologies.
Existing scholarly publications, while excellent within their specialized domains, often lack the breadth to accommodate research that operates at the intersection of multiple engineering fields, emerging technologies, and societal imperatives. Engineering Frontiers was established to fill this critical gap. We welcome contributions spanning the full spectrum from fundamental research with clear application pathways to experimental validation studies, computational modeling frameworks, and system integration methodologies. Our scope encompasses diverse areas including structural monitoring and prognostics, advanced manufacturing technologies, renewable energy systems, medical device innovation, autonomous systems, and beyond.
Our editorial philosophy rests on three foundational pillars: scientific excellence, engineering relevance, and innovation impact. We are committed to providing authors with a rigorous yet efficient peer review process that balances thorough evaluation with timely publication. Our distinguished editorial board brings together leading researchers and practitioners who ensure that published work meets the highest standards while remaining accessible to the broader engineering community. Through this approach, we aim to accelerate the translation of research insights into practical technologies that address pressing global needs, from advancing the United Nations Sustainable Development Goals to enhancing technological resilience across critical infrastructure systems.I extend my sincere appreciation to the researchers who have contributed to this inaugural issue, our dedicated editorial board members and reviewers who have ensured the quality and rigor of each manuscript, and our publishing partners who have provided essential support in establishing a platform that transcends disciplinary boundaries while fostering collaboration between academia, industry, and society.
We invite the global engineering research community to join us in this endeavor to advance knowledge at the frontiers of interdisciplinary engineering, where innovation meets impact and where scientific discovery translates into solutions for humanity's most pressing challenges.

39

Machine Learning-Based Fault Detection for UR3 Collaborative Robot: A Multimodal Data Analysis Approach

Shida Liu (Author)

摘要

Collaborative robots (cobots) are increasingly deployed in industrial environments, necessitating robust fault detection systems to ensure operational safety and efficiency. This study presents a comprehensive machine learning framework for fault detection in UR3 collaborative robots using multimodal sensor data. We analyzed a dataset comprising 7,409 samples with 20 features, including joint currents, temperatures, and velocities. Five machine learning algorithms were evaluated: Logistic Regression, Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (KNN), and Decision Tree. Our results demonstrate that KNN achieved the best overall performance with an accuracy of 94.42%, F1-score of 0.446, and AUC of 0.812. Feature importance analysis revealed that joint currents (J3, J2) and joint velocities (J5) are the most critical indicators for fault prediction. Dimensionality reduction techniques (PCA and t-SNE) confirmed distinct separability between normal and fault conditions. This work provides valuable insights for developing predictive maintenance systems in collaborative robotics applications.

39

Research on Short-Term Probability Forecasting Methods for Photovoltaic Power Generation Based on Multi-Source Data Fusion and Feature Engineering

Qiuhe Feng, Geyang Liu, Zhaoxi Zhao (Author)

摘要

The intermittent and uncertain nature of photovoltaic power generation poses significant challenges to grid stability. To enhance short-term forecasting accuracy and quantify uncertainty, this study proposes a probabilistic forecasting method for photovoltaic power generation that integrates multi-source data with feature engineering. Utilising operational data from a 50 kW actual photovoltaic power station combined with NASA meteorological records, a 29-dimensional feature system was constructed encompassing temporal characteristics, solar geometric features, physical statistical attributes, and meteorologically derived features. Employing a LightGBM quantile regression model, we achieve probabilistic forecasting of photovoltaic output for the next hour. Experimental results demonstrate excellent point prediction performance (RMSE = 7.323 kW, nRMSE = 0.085), with a P50 quantile loss of 2.4615 in probability forecasting, indicating effective uncertainty quantification capability. Feature importance analysis indicates that direct solar radiation, historical power lag terms, and solar elevation angle are key influencing factors. Although the P10–P90 forecast interval coverage (59.0%) reveals room for improvement in capturing extreme fluctuation events, this study provides a reliable data-driven solution for intelligent power system dispatch under high renewable energy penetration.

39

Task Affinity-Aware Scheduling for Multi-Core Edge Devices in Autonomous Vehicles

Zihe Hao (Author)

摘要

The evolution of autonomous driving systems toward cooperative perception has significantly intensified the computational workload on in-vehicle multi-core edge computing devices. Existing real-time scheduling strategies often fail to adequately consider task affinity when dynamically allocating heterogeneous core resources. When dealing with highly concurrent safety-critical tasks, this strategic limitation easily triggers frequent cross-core task migrations and severe cache invalidation. Consequently, this leads to a decline in overall system throughput and uncontrollable end-to-end response latency. In light of these complex challenges, this study draws cross-disciplinary inspiration from communication networks. Specifically, it adapts the scheduling logic that uses directed acyclic graphs to predictively avoid physical occlusion and reconstructs it into a mechanism for avoiding cache affinity disruption within the computational domain. Building upon this foundation, this paper proposes a forward-looking task affinity-aware scheduling framework. By deeply mining the temporal evolution characteristics of future computational workloads, this framework accurately maps the multidimensional decision space of multi-core resource allocation into a directed acyclic graph model. It continuously solves for a near-optimal scheduling path along the time series to maximize the net effective computational volume. Complex co-simulation tests preliminarily demonstrate that this mechanism effectively suppresses unnecessary computational jitter under extreme system load scenarios. Furthermore, it provides a solid underlying guarantee for enhancing the braking response safety of autonomous vehicles. This exploration provides a strong basis to further consider the broad theoretical boundaries and potential collaborative optimization directions for multi-core edge scheduling within future advanced autonomous driving architectures.

目录来源:OJS · 更新于:2026-11-18 06:17:10