Publication Search

Heating, Ventilating, and Air Conditioning (HVAC) systems often suffer from significant energy wastage due to their inability to adapt to real-time environmental changes, leading to high operational costs. Although Proportional-Integral-Derivative (PID) controllers are widely used for their simplicity and reliability, they struggle to handle the complex dynamics of modern environments, requiring advanced optimization to enhance efficiency. This study aims to optimize PID controllers by integrating the Queen Honey Bee Migration (QHBM) algorithm to improve HVAC performance, energy efficiency, and adaptability. The research method employs an experimental approach that compares the performance of conventional PID controllers with PID controllers optimized using the QHBM algorithm under dynamic environmental conditions. The results show that the PID-QHBM system significantly outperforms the conventional PID system, achieving a rise time of 0.2649 seconds and a settling time of 1.6874 seconds with an almost negligible steady-state error of 9.4991e-08. Although it experiences a slight overshoot of 16.3810%, the system stabilizes quickly and maintains the target temperature efficiently. In contrast, the conventional PID controller exhibits slower response characteristics, with a rise time of 1.3730 seconds, a settling time of 2.5144 seconds, and a larger steady-state error of 0.0361. This study demonstrates that integrating the QHBM algorithm into PID controllers provides a more effective solution for real-time temperature control, offering substantial improvements in energy efficiency and system performance. The findings contribute to advancing intelligent HVAC control systems that can better adapt to environmental variations while minimizing operational costs.

Hypertension is a major global health risk that requires accurate early detection, yet conventional methods struggle with complex and imbalanced health datasets. This study aims to optimize hypertension prediction using a Logistic Regression model integrated with Borderline-SMOTE to enhance recall and provide model transparency through SHAP (Shapley Additive Explanations). The method utilizes the BRFSS dataset, applying Borderline-SMOTE to address class imbalance at the decision boundary and XAI techniques for global and local interpretation. The findings show that the model achieved an accuracy of 0.719, an AUC of 0.800, and a significantly improved recall of 0.756. SHAP analysis identified age, high cholesterol, and BMI as the most influential risk factors, while waterfall plots successfully clarified individual risk extremes, ranging from 1.72% to 99.43% probability. These results imply that the proposed approach provides a sensitive and transparent screening tool for public health practitioners, effectively balancing statistical efficiency with clinical accountability.

This study aims to investigate the impact of social media advertising on clothing choices at Universitas Dinamika Bangsa Jambi students. In today's world, where many people, especially young people who frequently shop online, often struggle to accurately determine the quality of items. A quantitative approach was employed, with a survey as the primary method of data collection. A questionnaire was distributed online via Google Forms and successfully elicited responses from 102 active students who are also social media users. The sampling technique used was purposive sampling, with participants selected based on criteria that matched the focus of the study. The data were then processed using SmartPLS 4 software with the Partial Least Squares Structural Equation Modeling (PLS-SEM) method to test the relationship between variables. The main findings indicate that social media promotions have a strong positive influence on students' clothing purchasing decisions. This underscores the crucial role of targeted advertising strategies in the digital world in shaping consumer preferences. This research is expected to serve as a guide for clothing entrepreneurs in developing online marketing plans that better suit the tastes and needs of students as their target market.

Indonesia is a developing country located at the convergence of four tectonic plates, making it highly prone to natural disasters such as earthquakes, tsunamis, landslides, and volcanic eruptions. These frequent disasters highlight the critical need for reliable electricity during emergencies. However, disaster-affected areas often struggle to restore power due to accessibility issues. To address this, alternative energy sources are needed, and Solar Power Plants (PLTS) offer a practical solution. PLTS are easy to implement, depend only on sunlight, and provide clean energy with low carbon emissions. Under clear skies, solar radiation can reach 1,000 Watts per square meter, making it a powerful energy source. Additionally, PLTS systems are adaptable and can be deployed in various formats, including mobile units. This study focuses on designing a Mobile PLTS to support BASARNAS operations in disaster zones. Data collection was conducted using resources from BNPB, BMKG, BASARNAS, and NASA. The analysis includes the geographical characteristics of the site, solar radiation intensity, and the operational dimensions of the BASARNAS Mobile Truck. The study aims to determine the suitable system specifications and estimate the energy production capacity of the Mobile PLTS. The proposed design features 20 solar panels, each with a capacity of 300 Wp, producing an average of 27.70 kWh per day. It also includes 16 batteries for energy storage. The remaining space in the truck can be used for transporting logistics or essential tools, making it a multifunctional unit ideal for disaster response scenarios.

This study explores the integration of hybrid AI control models, combining reinforcement learning (RL) and robust adaptive control, to improve the adaptability, performance, and stability of autonomous manufacturing systems. Traditional control systems, while effective under stable conditions, often struggle to cope with disturbances and varying production demands. Hybrid AI models, which integrate classical control methods such as Proportional Integral Derivative (PID) with machine learning techniques like RL, deep Q-networks (DQN), and deep deterministic policy gradient (DDPG), enhance decision-making capabilities in dynamic production environments. The study develops a hybrid RL robust control framework and tests it in both simulation and real-world scenarios. Performance metrics, including production efficiency, system stability, and adaptability, are assessed under various disturbance conditions, such as machine failures and fluctuating demands. The hybrid model significantly outperforms traditional PID control in terms of efficiency and stability, demonstrating faster convergence and better adaptability in dynamic environments. Statistical analysis confirms the superiority of the hybrid system over standalone RL models and traditional PID control. This model’s scalability and adaptability make it a promising solution for Industry 4.0 applications, addressing key challenges in real-world manufacturing systems by ensuring computational efficiency and the ability to manage large-scale data. The findings contribute to the development of more robust and efficient control strategies for autonomous manufacturing systems in uncertain environments.