Energy storage systems (ESS) play a central role in renewable energy integration, grid reliability, and the transition toward low-carbon energy systems. In Malaysia, however, the indicators used to evaluate ESS remain fragmented, limiting comparison across technologies and weakening the evidence available for investment, policy, and sustainable supply chain decisions. This study investigates how ESS performance has been evaluated in the Malaysian energy transition and develops a structured framework for linking engineering performance with sustainable supply chain management (SSCM). A systematic review of 40 eligible studies was conducted using bibliometric mapping and thematic analysis. The reported indicators were identified, coded, and classified into technical, economic, operational, and policy/environmental dimensions. The results showed that capacity and sizing were the most frequently reported indicators, followed by renewable energy integration and system reliability or availability. Battery-based systems dominated the reviewed literature, particularly in photovoltaic (PV)-coupled applications, whereas long-duration storage, grid-scale services, lifecycle assessment, and end-of-life considerations received limited attention. Although levelized cost of energy and net present cost were commonly reported, none of the retained studies explicitly evaluated the levelized cost of storage (LCOS). The findings indicate that current assessment practices remain concentrated on project-level technical and financial performance and provide insufficient support for evaluating material sourcing, lifecycle impacts, regulatory conditions, and supply chain resilience. The proposed framework connects ESS performance evaluation with technology selection, investment appraisal, supplier assessment, environmental management, and policy planning. It provides a systematic basis for developing national performance benchmarks and supports more consistent ESS decision-making in Malaysia and other Association of Southeast Asian Nations (ASEAN) energy systems.
Mobile photovoltaic (PV) power systems provide a flexible electricity supply for remote locations, emergency operations, and other off-grid applications. Their practical operation requires continuous assessment of power conversion performance under changing environmental conditions. This study investigates the efficiency, power output, and operational reliability of a mobile PV system through an Internet of Things (IoT)-enabled monitoring platform and a multilayer perceptron (MLP) model. Solar irradiance, panel temperature, voltage, current, power, and battery state of charge (SOC) were recorded under outdoor operating conditions, yielding approximately 1,600 observations. An MLP with two hidden layers was trained using the Levenberg–Marquardt algorithm, and the data were divided into training and testing subsets at a ratio of 80:20. Operational reliability was evaluated by comparing measured and predicted power outputs against statistically defined control limits. The PV panel achieved an average operating efficiency of approximately 15%, whereas the efficiency of the solar charge controller (SCC) reached 60%. For the normalized dataset, the MLP produced mean squared error (MSE) values of 0.002402 and 0.001951, root mean squared error (RMSE) values of 0.049012 and 0.044173, mean absolute error (MAE) values of 0.033774 and 0.027760, and $R^2$ values of 0.964491 and 0.970248 for PV and controller power, respectively. The predicted outputs remained within the established control limits throughout the observation period. These findings indicate that the proposed framework can support real-time power-performance assessment and the early identification of abnormal operating conditions in mobile PV systems.
Agentic data pipelines, in which large language models select and invoke tools through the Model Context Protocol, consume tool outputs, and iteratively execute multi-step analytical or operational workflows, are increasingly being deployed in production environments. However, the observability infrastructure required to diagnose failures in such systems remains underdeveloped. Conventional distributed tracing effectively captures service-to-service execution but often represents large language model invocations as opaque spans and fails to preserve causal relationships across large language model-tool boundaries. Consequently, incident diagnosis can require an agent's execution trajectory to be reconstructed manually from chronologically ordered spans. To address this limitation, causal span linking across large language model and tool invocations was defined as a first-class observability primitive for agentic data pipelines. Hops-to-root-cause was introduced as the directed acyclic graph distance between a symptom span—the earliest span tagged error=true—and the identified causal span, with deterministic tie-breaking applied. The proposed approach was evaluated on a synthetic corpus comprising 20 incidents generated using a fully disclosed construction protocol. Compared with a flat-span baseline, causally linked tracing reduced the mean hops-to-root-cause from 6.4 to 1.5, corresponding to a reduction by a factor of 4.3. The greatest improvements were observed for incidents involving multi-hop tool chains. Causal span linking was further complemented by deterministic replay and by a human-artificial intelligence collaborative diagnostic workflow. Together, causal span linking, deterministic replay, and human-artificial intelligence collaborative diagnosis were established as complementary observability primitives for improving the reproducibility, interpretability, and efficiency of root-cause analysis in agentic data pipelines.
Immersive technologies are increasingly used by cultural institutions to create context-sensitive visitor experiences, yet conventional media branding pipelines rely largely on predefined visual assets and provide limited support for real-time adaptation. This study investigates how generative artificial intelligence (AI) can be integrated into an adaptive systems architecture while preserving institutional visual identity. A mixed-method design was employed, comprising an analysis of immersive branding pipelines, case studies of five cultural institutions, the development of two prototype application scenarios, and an evaluation by nine experts. The proposed architecture connected contextual data acquisition, generative processing, constraint validation, immersive rendering, and user feedback within a closed-loop workflow. A Validator module was introduced to examine generated outputs against predefined color and geometric constraints and to initiate regeneration or fallback procedures when violations were detected. The case analysis produced a mean adaptivity score of 4.2 out of 10 for the existing implementations. Expert evaluation of the proposed architecture yielded mean scores of 4.78 for personalization, 4.56 for visual identity flexibility, and 3.89 for brand consistency. Generation latency ranged from 1.2 to 1.8 s in the augmented reality (AR) scenario and from 2.5 to 4.0 s in the virtual reality (VR) scenario. The findings indicate that generative AI can be incorporated into a feedback-controlled branding pipeline without removing deterministic control over core visual elements. The proposed architecture provides a systems engineering basis for coordinating content generation, identity validation, and immersive delivery, while the observed latency and limited evaluation sample identify priorities for edge deployment and larger-scale experimental validation.
Multi-tier supply chains are exposed to operational disruptions that can spread through interconnected supplier–customer relationships. Existing risk assessments often rank individual firms or links without considering whether these relationships form continuous routes of concentrated vulnerability. This study investigates critical-path identification as a decision-analytics problem in which relationship-level performance and network structure are considered jointly. A two-phase framework was developed by integrating the Criteria Importance Through Intercriteria Correlation (CRITIC) method with evolutionary path optimization. In the first phase, CRITIC was used to derive objective weights for 14 operational performance criteria and to calculate a criticality index for each supplier–customer link. In the second phase, two optimization models were formulated to identify the path with the highest cumulative criticality and the path with the highest average link criticality subject to a minimum path length. The framework was applied to an automotive supply chain consisting of 19 enterprises and 35 directed links. The cumulative model identified a five-link path with a total criticality of 3.274 and an average criticality of 0.655, whereas the average-criticality model identified a four-link path with a total criticality of 2.738 and an average of 0.684. Both models selected the same initial relationship but produced different subsequent routes. The results indicate that link-level rankings alone cannot identify the most critical continuous route because path selection also depends on connectivity, topological position, and the optimization objective. The framework provides a reproducible basis for prioritizing supplier relationships, directing monitoring resources, and selecting risk-mitigation measures across multi-tier supply networks.
Information asymmetry remains a fundamental constraint on the allocation of external finance, as the organisational capabilities and internal quality of firms cannot be fully observed by external financiers. Although managerial quality, human capital, innovation, digitalisation, relational networks and financial transparency have each been linked to financing outcomes, their interrelated nature has received considerably less attention. To address this gap, a unified theoretical framework is developed in which these observable organisational attributes are conceptualised as indicators of a latent construct, termed Organisational Quality (OQ). Drawing on information asymmetry theory and signalling theory, the framework is tested using longitudinal data from 2,017 Vietnamese manufacturing small and medium-sized enterprises (SMEs), comprising 6,051 firm-year observations from the SMEs Survey. OQ is operationalised as a reflective latent construct within a structural equation modelling (SEM) framework, through which its associations with access to external finance, productive investment and firm performance are examined. Strong empirical support is obtained for the proposed framework. Higher OQ is found to be positively associated with access to external finance, and this association is significantly stronger under conditions of greater information asymmetry. Access to external finance is, in turn, positively associated with productive investment, while productive investment is positively associated with firm performance. The mediation results further indicate that OQ is associated with firm performance through both direct and indirect pathways, with a sequential pathway operating through improved access to external finance and subsequent productive investment. These relationships remain stable across alternative measurement approaches, model specifications and measures of firm performance. The findings provide three principal contributions. First, organisational characteristics that have traditionally been examined separately are integrated into a common latent organisational dimension. Second, OQ is operationalised and empirically validated as a reflective latent construct within an integrated SEM framework. Third, evidence is provided that OQ is associated with improved access to external finance, greater productive investment and enhanced firm performance, while its financing-related association becomes more pronounced as information asymmetry increases.
Intelligent transportation systems increasingly rely on connected and autonomous platforms operating across road, rail, low-altitude, and marine environments. These systems must accommodate nonlinear and time-varying dynamics, environmental uncertainty, communication limitations, safety requirements, and tightly coupled operational constraints. Model predictive control (MPC) is well suited to such conditions because it combines future-state prediction, constrained optimization, and closed-loop correction within a receding-horizon framework. This review examines the theoretical foundations, major formulations, and transportation applications of MPC across four domains: ground vehicles and traffic networks, railway systems, low-altitude unmanned aerial transportation, and marine autonomous systems. The literature was organized according to transportation mode and application, with attention given to prediction models, control objectives, operational constraints, uncertainty treatment, computational requirements, and validation methods. The review showed that ground transportation research placed greater emphasis on vehicle motion control, connected-vehicle coordination, traffic signal optimization, and network regulation. Railway applications focused mainly on train regulation, virtual coupling (VC), scheduling, energy-efficient operation, and maglev control. Low-altitude studies addressed trajectory tracking, obstacle avoidance, multi-unmanned aerial vehicle (UAV) coordination, and learning-based prediction, whereas marine studies concentrated on underactuated motion, environmental disturbances, collision avoidance, navigation rules, and surface–underwater cooperation. Across these domains, model uncertainty, online computational burden, conflicting control objectives, coupled safety constraints, and limited real-world validation remained the principal obstacles to wider deployment. The findings indicate that MPC architectures must be adapted to the dynamics, operational environment, and communication conditions of each transportation mode. This review clarifies the common and domain-specific requirements of MPC in multimodal intelligent transportation and identifies the technical issues that require further theoretical and experimental study.
Vietnam’s post-pandemic tourism recovery raises the question of how renewed visitor growth relates to economic value, climate vulnerability, and mobility-related carbon performance. This study assesses Vietnam’s tourism recovery under climate change and the Net Zero transition using secondary data for 2019–2023. It applies descriptive analysis, recovery indices, system-level emission-to-activity ratios, passenger-aviation Tapio decoupling analysis, and a national-transport system-level proxy elasticity for 2022–2023. National transport and passenger-aviation emissions are treated as system-level proxies for mobility-related carbon pressure and do not constitute a tourism emissions inventory. The results indicate a scale–value–carbon mismatch. In 2023, total tourist volume reached 117.3% of its 2019 level, whereas nominal tourism revenue and average nominal revenue per reported tourist visit reached 89.8% and 76.5%, respectively. Contextual climate evidence indicates vulnerability channels affecting destinations, infrastructure, transport connectivity, business operations, and service continuity. National transport and passenger-aviation emissions reached 104.1% and 93.8% of their 2019 levels. During 2022–2023, national transport emissions grew more slowly than measured passenger activity, while passenger aviation exhibited weak decoupling. Absolute emissions nevertheless increased in both systems. The results indicate short-run relative improvement in emissions performance rather than structural decarbonisation. Three policy priorities emerge: strengthening tourism carbon accounting, integrating climate adaptation into destination and transport planning, and coordinating measures to reduce mobility-related emissions. The diagnostic framework may also be relevant to tourism transitions in emerging economies facing similar climate, mobility, and data constraints.
For Authors
For Reviewers
For Editors
Open Access
