ABSTRACT Reliability‐Centred Maintenance (RCM) is an economical maintenance approach that places more emphasis on sustaining system functionality than just fixing individual hardware parts. Failure Mode and Effects Analysis (FMEA), one of the main techniques in RCM, is used to determine possible failure modes, their underlying causes, and how they affect system performance. This study identifies the greatest risk that affects the system by using a higher risk priority number (RPN) and clearly illustrates the maintenance plan for each failure scenario. By putting in place a systematic RCM policy, the study seeks to reduce these risks. The results show that, in terms of the impact of failure, the machines in the spinning portion are the most crucial. By lowering risk through a focused FMEA approach, the RCM methodology's implementation shows notable economic benefits. The suggested approach improves maintenance efficiency by concentrating on the most important 20% of failure causes. According to the findings, implementing RCM resulted in a 31.5% decrease in labour expenses. Additionally, compared to the case company Bahir Dar Textile Share Company's (BDTSC) present run‐to‐failure maintenance policy, the proposed maintenance plan reduced the yearly spare parts costs by 28.12%. The results highlight RCM's potential for enhancing maintenance plans, cutting expenses, and raising overall operational effectiveness in the textile industry.

ABSTRACT This study centers on a 45# mineral oil and FR3 natural ester transformer oil blend. Under extremely cold conditions, it explores how varying moisture content impacts the breakdown characteristics of this mixture and examines the internal factors related to breakdown voltage. Moisture absorption experiments, breakdown tests, and computational simulations of moisture migration trajectories within oil matrices were conducted, with data analysis for temperatures below 0°C using the Weibull distribution. Results indicate that at a fixed oil mixture ratio, reducing moisture content greatly enhances the insulation oil's insulation performance. Around –30°C, increasing the natural ester proportion in the mix effectively raises the breakdown voltage. Under low—temperature conditions (0°C and below), when moisture content is low, increasing the natural ester share most markedly improves the insulation ability. However, as moisture content rises, this enhancement effect diminishes. Through simulation, the movement of moisture droplets in the transformer oil was analysed. It was found that they migrate to high‐electric‐field‐strength areas and gather, stretch, and connect the two poles. Further research shows that the movement speed of free water in mineral oil is about two orders of magnitude faster than in natural ester, and this speed difference grows with increasing voltage.

ABSTRACT The ChatGPT application is widely utilised across diverse research domains; however, its adoption within educational libraries remains limited. This study investigates the integration of ChatGPT into online academic libraries and introduces a ChatGPT‐based framework designed to enhance user experience and service delivery. The framework was evaluated using two empirical metrics: the user satisfaction score (USS) and the resource engagement rate (RER). The evaluation results indicate that ChatGPT prompts can generate realistic scenarios and facilitate active role‐playing, thereby improving the quality and variety of library services. Empirical data show that the proposed approach achieved high USS values, ranging from 4.4 to 4.8 out of 5, while RER increased to 79.5% between January and May 2023. These findings suggest that ChatGPT has considerable potential to strengthen educational library services. Specifically, its integration supports personalised assistance, timely responses and access to subject‐specific expertise. Moreover, AI‐driven Chatbots can curate digital resources, design tailored learning pathways and promote self‐directed learning. Beyond individual support, ChatGPT contributes to collaborative learning environments and ensures users’ access to current and diverse information resources.

ABSTRACT Origami‐inspired robots hold significant potential in flexible systems, yet their operational speeds remain constrained by the inherent trade‐off between structural compliance and actuation efficiency. This study presents a novel approach by exploiting the transient energy release characteristics of bistable Kresling origami to achieve rapid locomotion. Through parametric modelling of the critical bistable transition, we establish quantitative relationships between geometric parameters and dynamic performance metrics, revealing that optimised configurations enable ultra‐fast rebound within 60 ms. The robot integrates a cable‐driven crank mechanism that synchronises motor rotation with periodic linear actuation, achieving precise control over energy storage–release cycles. A bioinspired differential friction foot design with programmable anchoring sequences facilitates earthworm‐like peristaltic motion. Experimental results demonstrate exceptional crawling speeds up to 375 mm/s (3–4 body lengths/s) across multiple substrates (ceramic, wood, PVC). This work provides a paradigm‐shifting framework for reconciling speed and flexibility in soft robotics, with direct implications for applications requiring rapid reconfiguration in confined environments.

ABSTRACT As network‐based services continue to expand and cyber threats grow increasingly sophisticated, the deployment of effective network intrusion detection systems (NIDS) has become crucial for the protection of sensitive data and the preservation of digital infrastructure integrity. These systems are indispensable in ensuring the security and reliability of digital environments. This study presents an all‐encompassing approach that integrates feature selection, ensemble learning, clustering, and diverse analytical methods to improve the precision and effectiveness of network intrusion detection systems. Using data sets that have been preprocessed and classified using the Random Forest and XGBoost algorithms, we extract key features and categorize key variables using K‐means clustering, thus constructing a robust feature set for intrusion detection. Our experimental results demonstrate high precision in detecting benign and malicious activities using the CIC‐IDS‐17 and UNSW‐NB15 datasets. These outcomes underscore the efficacy of our proposed approach in monitoring network traffic, effectively identifying threats, and preventing potential intrusions.

ABSTRACT A simple and derivative‐free root‐finding method based on geometric reflection and box construction is introduced. The proposed reflected box method achieves quadratic convergence like Newton's method, but operates without requiring derivatives. Its arithmetic simplicity makes the method naturally amenable to fixed‐point and embedded implementation, though this paper focuses on floating‐point evaluation to establish baseline performance. Numerical experiments confirm competitive performance against Brent's and Steffensen's methods, with fewer iterations, simpler arithmetic and comparable residual accuracy.

ABSTRACT This study develops a numerical algorithm for unsteady fluid flow for short and long hydraulic systems, including various hydraulic devices in the system, through scientific searches for new ways to protect long pipeline systems from sudden pressure surges and prolonged pipeline exposure to high pressure. One of the methods is the localisation of high‐pressure zones with the help of oscillating circuits. Numerical experiments show that the frequency of pressure oscillations in them increased tens of times, making it possible to dampen the energy of pressure waves more effectively with the help of pressure stabilisers of smaller volumes. The amplitude of pressure oscillations decreases rapidly due to energy loss from the friction of the liquid against the pipeline walls. Studies have been carried out for oscillatory circuits with the introduction of pipeline sections with low‐pressure wave propagation velocity. These calculations reveal additional possibilities for pressure reduction.

ABSTRACT Chronic diseases are long‐lasting diseases that last for three months or longer. Remote Patient Monitoring Systems (RPMS) contain smart and advanced technologies such as Artificial Intelligence (AI) and the Internet of Things (IoT) for monitoring patients remotely to enhance patient experience and comfort while reducing hospital visits. Current RPMS architecture often lacks the proper integration of AI and IoT, concentrating on data collection and data transmission without focusing on real‐time processing, predictive analytics and scalability. This research aims to propose an AIoT phase‐driven architecture for RPMS, including evaluating the requirements for managing and monitoring chronic diseases remotely, effectively offering a structured, scalable and intelligent architecture that incorporates IoT for the data collection tasks and AI for real‐time analytics and decision‐making. AIoT phase‐driven architecture is a unique and well‐structured architecture for RPMS, especially by integrating AI and IoT technologies, allowing real‐time monitoring and personalised treatment plans with accurate and precise disease classification. To evaluate and validate the proposed AloT phase‐driven architecture for RPMS, a survey has been conducted with domain experts, which confirms the proposed architecture and the evaluation of the requirements for scalability, practicality and potential to improve chronic disease management with strong support for its phase‐driven approach based on AIoT. The proposed AIoT phase‐driven architecture provides a robust and scalable solution for chronic disease monitoring, enhancing overall patient outcomes and reducing healthcare burdens.

ABSTRACT Accurate assessment of substation equipment conditions is crucial for ensuring the reliable operation of power systems. Traditional methods rely on centralised training, which requires extensive labelled data often owned by different electricity companies. Driven by the need for privacy protection and competitive advantage, such data cannot be shared, creating significant barriers to effective model development. To address these challenges, this paper proposes a federated learning‐based Transformer model (FL‐TM). The Transformer network is utilised to extract global features from equipment images captured by inspection robots, while a multi‐layer perceptron (MLP) classifier performs classification tasks. To minimise communication overhead, only the MLP is updated locally, while the Transformer network remains frozen during training. The proposed FL‐TM framework incorporates a knowledge distillation‐based mechanism, where soft labels generated by edge nodes are aggregated at the cloud server to optimise the global model while preserving data privacy. Experimental results show that, in distributed and privacy‐sensitive environments, the proposed method enhances recognition accuracy and achieves lower communication costs.

ABSTRACT Traditional duty‐cycle or phase‐shift modulation is inadequate for achieving a balanced current‐stress distribution during the startup of full‐bridge DC/DC converters. As converter power capacity increases, these limitations not only restrict the startup speed but also elevate the risk of device damage. This letter introduces a high‐power full‐bridge series‐resonant converter (SRC) based on an integrated gate‐commutated thyristor (IGCT) designed for photovoltaic DC collection. A novel phase‐shift modulation is then proposed to balance the current‐stress distribution across all semiconductor devices during startup. Additionally, the beneficial impact of the secondary‐side buffer capacitors on the startup process is discussed. Experimental results are presented to validate the theoretical analysis and demonstrate the effectiveness of the proposed modulation.