With the increasing proportion of renewable energy sources, the output characteristics of wind and solar power in power systems have become increasingly stochastic. This phenomenon not only affects clean energy utilization efficiency but also poses significant challenges to the safety and stability of traditional power grids. In these conditions, energy storage devices have become essential for reducing power fluctuations from renewable sources, and they are crucial for source-storage joint dispatching and dynamic control. Through literature review and comprehensive analysis, this study investigates characterization methods and quantification approaches for renewable energy uncertainty within source-storage coordination frameworks, explores optimization model design philosophies and related strategies, and evaluates current mainstream control technology trends. The uncertainty assessment system, established using parameters such as prediction deviations and fluctuation amplitudes, demonstrates greater scientific rigor than conventional coarse scenario segmentation methods. Multi-objective function-based collaborative planning approaches are gradually replacing single-objective modeling paradigms. Regarding real-time response performance improvement techniques, model predictive control (MPC) integrated with reinforcement learning algorithms has become a new research focus. This study aims to clarify academic frameworks in relevant fields and provide theoretical foundations and technical support for practical applications of source-storage coordination mechanisms in high-renewable-energy environments.

Wide bandgap semiconductors, particularly gallium nitride (GaN) and silicon carbide (SiC), are critical materials for next-generation electronic and high-frequency devices. However, material defects and interface states significantly degrade device performance and reliability. This study presents an integrated multiscale computational framework combining density functional theory, molecular dynamics simulations, and machine learning approaches to analyze defect formation and interface optimization in wide-bandgap semiconductor systems. Key findings include: (1) identification of dominant dislocation configurations in GaN heteroepitaxy and their impact on electron mobility; (2) development of a graph neural network model that predicts defect formation energies with an accuracy exceeding 85%; and (3) optimization of a SiC/SiO₂ interface passivation scheme that reduces interface state density by 60%. The proposed framework bridges atomic-scale defects to device-level performance, providing actionable insights for material synthesis and device fabrication.

With the rapid development of information technology, smart port and shipping has become an inevitable trend in the development of port and shipping field. This paper focuses on the innovation and development of logistics management under the background of smart port and shipping, and expounds its important position in economic development through an overview of port and shipping. In-depth analysis of the application and development of intelligent port and shipping information technology, including automated terminal technology, blockchain technology and big data technology in the port and shipping field, taking Yangshan Port as an example to discuss how these technologies can improve the efficiency of logistics management, reduce costs, enhance safety and transparency, and provide theoretical support and practical guidance for the reform of port and shipping logistics management.

The sharp increase in urban vehicle ownership has triggered severe traffic congestion and low road network operational efficiency, making intelligent transportation systems (ITS) a core technical means to optimize urban traffic management. Nevertheless, existing studies have obvious deficiencies: there is a lack of systematic collation and comparative analysis of ITS practical applications across diverse urban scenarios, and the exploration of developmental constraints is fragmented without integrating inherent correlations among industrial standards, equipment compatibility and operational costs. This paper adopts literature research, case analysis and inductive methods to select typical urban cases in Indonesia, Poland and Malaysia, exploring the practical effects of ITS on signal timing optimization, short-term traffic flow prediction and multi-intersection collaborative control under traffic disturbances. The results indicate that simulation modeling, regression algorithms and deep reinforcement learning can effectively reduce traffic queue length and travel delay, and improve traffic prediction accuracy and road network operation efficiency. This study further identifies three systemic bottlenecks restricting ITS large-scale promotion: poor adaptability to extreme weather and unexpected traffic incidents, insufficient standardization and scalability for city-wide deployment, as well as data privacy risks and user psychological resistance. It also clarifies research limitations and proposes future research directions, aiming to provide theoretical support and practical references for urban intelligent traffic governance and bottleneck breakthroughs.

Magnetorheological fluid is a kind of intelligent material that is widely used in fields such as damping and vibration reduction, transmission control, sealing and precision machining, and can rapidly undergo rheological transformation under an external magnetic field. However, the large density difference between magnetic particles and carrier fluid easily leads to problems such as particle sedimentation, agglomeration and poor redispersibility, thus limiting the application effect of magnetorheological fluids. This paper reviews the preparation methods of magnetorheological fluids with high dispersion stability, focusing on methods including additive modification, addition of magnetic nanoparticles, surface coating of carbonyl iron powder, introduction of two-dimensional fiber materials and optimization of base fluid. Studies have shown that methods such as using high-viscosity carrier fluid, adding nanomaterials and coating particles to form core-shell structures can effectively improve the dispersion stability, yield stress and redispersibility of magnetorheological fluids. The current development status of the preparation of magnetorheological fluids with high dispersion stability is summarized, and the future directions of multifunctional composite materials, green and environmentally friendly base fluids and intelligent preparation are prospected, so as to provide references for the research and application of high-performance magnetorheological fluids.

The Architecture, Engineering, and Construction (AEC) industry has historically grappled with severe, chronic cost overruns, rendering traditional deterministic estimation techniques obsolete for modern megaprojects. A detailed review is used to examine how predictive machine learning (ML) algorithms have developed over time and how they compare to one another in terms of performance for predicting construction costs. The review traces the technological advancements made over time and how this has affected the development of predictive models. It also looks at how predictive models have traditionally relied on first-generation linear regressions and kernel methods, which do not capture the complex nature of the construction process, as compared to second-generation tree ensembles (which are an improvement over first-generation methods). Specifically, the overwhelming superiority of Random Forest, XGBoost, and, notably, CatBoost in processing the chaotic, high-dimensional, and heavily categorical data typical of the AEC sector is analyzed in depth. Furthermore, this review explores the third-generation paradigm shift towards Deep Learning (DL). As projects increasingly generate complex, multi-level temporal data, Long Short-Term Memory (LSTM) networks have emerged as indispensable tools for modeling time-evolving dynamics, such as macroeconomic inflation and sequential supply chain disruptions. Using an evidence-based roadmap to select optimal predictive architectures based on data topology, this review enables AEC stakeholders to transition from reactive cost accounting to proactive financial risk management by systematically comparing these algorithmic generations.

More than 80 percent of the global goods are transported by sea. Traditional ships use the heavy oil and diesel as the fuel, and cause environmental pollution. In order to deal with this problem, the International Maritime Organization has set the targets for the shipping emission reduction. China is also working towards the goals of carbon peaking and carbon neutrality. The marine power systems should turn to the green energy.This paper studies four new fuels for marine internal combustion engines: liquefied natural gas (LNG), methanol, hydrogen and ammonia. This paper explains the features, advantages and disadvantages of each fuel. And then, it makes a full comparison from five aspects: the environment protection, the technology, the storage and transportation, the safety and the commercial use.The research results indicate that LNG is a transitional fuel with mature technology and high safety. Methanol is easy to use and will be widely promoted in the near future. Hydrogen and ammonia can realize zero carbon emissions. Hydrogen fit inland and short-distance ship, and ammonia works better for ocean-going vessels. The marine new fuels will develop in diverse ways in the future.

With the dual carbon goals, the environmental impact and energy consumption of the construction stage have gradually received more attention. Although the smart construction site system has been introduced, data management is not complete; some monitoring systems are not integrated, and real-time Building Information Modelling (BIM) data application is still lacking. This paper introduces a BIM and Internet of Things (IoT) integrated framework to combine BIM models with on-site IoT devices for environmental monitoring and energy consumption management. Based on the proposed approach, the system will be able to monitor the environment and energy consumption in a single window and support BIM-based data management. Collaborative optimization strategies have been introduced, such as threshold-triggered dust suppression, dynamic energy scheduling according to construction progress, and the combination of BIM and Geographic Information System (GIS), edge computing and digital twins for improved decision-making. Based on the reviewed studies, the new architecture can facilitate more immediate sharing of data, enhance coordination among environmental control and energy management systems, and provide supporting data for on-site operations. The existing method is a fixed-threshold type, so it has deficiencies in the speed of response, and the new framework provides support for intelligent construction management. In the future, deep learning technology will be applied to improve the accuracy of emission prediction and extend the scope of the framework to cover the whole life cycle of emissions.

This study focuses on a proposed composite public space in Xiwusutu Village, Hohhot, and discusses the regional expression, construction adaptability and functional integration of rammed earth in rural public space. Drawing upon field surveys, questionnaire research and comparative case analysis, this paper systematically sorts out rammed-earth residential architecture distributed across the southern Yinshan foothills and Hohhot-Baotou Basin, and links this building material to local temple culture, integrated farming-nomadic collective memory, ancient apricot orchard landscapes and the spatial logic of loess terraced landforms. The design reorganizes the site's idle plots into a multifunctional public space integrating folk culture exhibitions, intangible cultural heritage workshops, study classrooms, visitor rest areas and activity spaces for local villagers. The study suggests that rammed earth can strengthen local identity and contribute to passive environmental performance, while its practical implementation still depends on soil testing, local construction training and detailed technical development.

Construction accidents occur frequently, and traditional on-site management methods lack adequate risk foresight. Manual supervision has a narrow scope and a slow response, and thus cannot meet the safety requirements in a changing construction environment. Smart construction sites use digital technology to promote the improvement of construction safety. Most of the current research has focused on individual technologies or functions of platforms and has failed to systematically integrate the analysis of the technological framework, application results and practical problems. Based on the above analysis, this paper establishes the core concept and structural components of a smart construction site as a new safety management model, which consists of three levels: an all-encompassing management platform, Building Information Modeling (BIM) technology, and a six-tier Internet of Things (IoT) control system. Based on the above foundation, the New Shandong Provincial Science and Technology Museum project is taken as an example, and the deployment method and working logic of a smart construction site in specific construction scenarios across four dimensions are analyzed. The results show that this model has achieved the following management objectives: shortened the schedule, reduced material costs, and decreased quality rework. Finally, the paper also analyzes the practical constraints of smart construction site promotion, such as unstable equipment, inconsistent data standards and an unreasonable safety investment structure, and puts forward targeted improvement suggestions. By systematically examining the technical framework, application effects and improvement directions of smart construction sites, this paper offers references for the digitalization of construction project safety management.