Background: There is an ongoing debate on the topic of Intermittent Fasting (IF) and whether this dieting method is either beneficial or detrimental for working individuals who decide to abstain from food at work.
Aims: To determine whether two consecutive days of fasting have an influence on work performance, distraction, and mood at work. Social support, fasting experience, age, and gender are also observed as confounding variables.
Design: This was a quasi-experimental, within-participants study, with repeated measures design (three measurement points, one per week), and a replication of Appleton and Baker‘s (2015) study, with some alternations.
Method: 33 participants (13 males and 20 females) aged between 21-54 fasted for two consecutive days at work. Measures were taken one week preceding, during, and after the fasting intervention.
Results: MANOVA reported a statistically insignificant difference of IF on work performance, distraction, and mood across the three weeks study.
Conclusion: Two consecutive days of IF did not influence cognitive performance at work in the given sample size. These findings can contribute to the field of Occupational Psychology by informing cohorts that the choice of abstention from food at work does not influence individual work performance, mood, and distraction at work.
Keywords: intermittent fasting, work performance, distraction, mood.

Existing fracture testing standards mainly focus on Mode I and Mode II fracture behaviors, with specimens being costly and complex to manufacture, while the Edge Notched Disc Bend (ENDB) specimen, due to its simple geometry and low cost, has been proven to be an effective method for assessing fracture toughness in pure modes (I, II, III) and mixed modes (I/II, I/III, I/II/III); this paper reviews the current research progress on ENDB specimens, covering their development history, basic principles, fracture toughness calculation formulas, loading modes, geometry factor calculations, and the evaluation of fracture characteristics and performance in polymers, composites, and single materials, emphasizing the flexibility of ENDB specimens in testing both pure and mixed fracture modes, summarizing the integration of ENDB experiments with various prediction criteria, models, other testing techniques, and different operational environments, identifying challenges in four areas—standardization, advanced testing techniques, testing of heterogeneous materials, and testing under long-term coupled chemical and physical environments—and providing recommendations such as the introduction of new testing technologies, optimization of formulations, improvement of prediction criteria, and integration with finite element analysis, offering references for the future development of ENDB testing and promoting the advancement of fracture mechanics research in pure and mixed modes.
Keywords: Edge Notched Disc Bend, Fracture Toughness, Experimental Methodology, Numerical Simulation, Standardization

As an important branch in the field of artificial intelligence, the research and development of AI agents have received extensive attention. This article elaborates on the background, history, technology, applications, challenges, and future trends of AI agents. Firstly, the research background of AI agents is analyzed from the aspects of computing power improvement, big data accumulation, algorithm and theoretical progress, social and economic needs, policy and financial support, and scientific exploration drive. Secondly, the development process of AI agents from the 50s of the 20th century to the present is reviewed, and the key technologies and representative achievements at different stages are summarized. Then, the core technologies of AI agents were introduced in detail, including machine learning, deep learning, natural language processing, computer vision, etc. Then, the application practices of AI agents in various fields, such as industrial automation, medical healthcare, financial analysis, intelligent transportation, etc., are discussed. In addition, this paper analyzes the challenges faced by AI agents in the development process, such as ethics, data security, and technical bottlenecks. Finally, the future development trend of AI agents is prospected, including technology integration, general intelligence, ethics and regulations, etc. The purpose of this paper is to provide comprehensive theoretical reference and practical guidance for researchers and practitioners of AI agents, and to promote the healthy development and application innovation of AI agent technology.
Keywords: AI agent; Background; Phylogeny; Technology; Apply; Challenge; Trend

This study addressed the challenges associated with routine power transformer (PT) maintenance strategies by using Random Forest (RF) model to predict PT maintenance. The goal was to reduce unexpected failures and unnecessary costs associated with both breakdown and preventive maintenance scenarios. Utilizing a dataset from several diagnostic tests conducted during breakdown and preventive maintenance, specifically the 3-phase and short circuit tests, the study aimed to predict the maintenance interval rate for proper scheduling of PT maintenance. The methodology formulated transformer maintenance as a supervised binary classification task, distinguishing between good operating conditions (Class 0) and conditions needing maintenance (Class 1). The RF model achieved an impressive 97% accuracy, demonstrating outstanding performance in predicting the maintenance needs of power transformers. The system architecture involved stages of collecting data, data preprocessing, feature selection, training of model, and maintenance prediction. The F1-score, precision, and recall metrics of the RF model illustrates very high performance, particularly in identifying transformers in good operating conditions. While plotting maintenance predictions against current dates, it was observed that regular maintenance check was required twice within a month, with a 14-day interval between these events. However, the usual annual preventive maintenance can still be done just to ensure proper working condition of all components of the power transformer. Additionally, comparative analysis revealed that the RF model is simpler to implement, requires minimal hyperparameter tuning, trains faster, and is more computationally efficient compared to other tree-based machine learning (ML) models like XGBoost, Light GBM, AdaBoost, and ANN. These advantages make RF a practical and reliable choice for real-world predictive maintenance applications.
Keywords: Machine learning (ML), Power transformer, Predictive maintenance, Random Forest (RF) classifier model, Transmission sub-station