Showing 4 results for Type of Study: Research
Dr Loke Kok Foong, Dr Wan Amizah Wan Jusoh, Dr Vellapandian Ponnusamy,
Volume 1, Issue 1 (9-2025)
Abstract
Comprehensive occupancy data from buildings may help energy management systems work better and save energy without sacrificing tenant comfort. Occupancy data is obtained by various methods, including high-precision instruments like thermal and optical cameras and specialized or environmental sensors like carbon dioxide (CO2) and passive infrared (PIR). While less accurate, the latter methods have drawn much interest since they are less expensive and invasive. This study's primary goal is to provide a unique prediction model that enhances the knowledge gleaned from sensor data by fusing evolutionary algorithms with machine learning (ML) models. To detect occupancy in an office setting, this study examines the effectiveness of four optimization algorithms: Multi-Verse Optimization (MVO), Teaching-Learning-Based Optimization (TLBO), Wind-Driven Optimization (WDO), and Whale Optimization Algorithm (WOA) when combined with Multilayer Perceptron (MLP). Among the noteworthy results is the strong performance of WDO-MLP, which shows remarkable AUC scores with little sensitivity to changes in swarm size. Furthermore, the best-performing approach is MVO combined with MLP, which shows better stability and discriminating ability at different swarm sizes. The practical implications of the research findings for implementing occupancy detection systems in smart buildings are noteworthy.
Comprehensive occupancy data from buildings may help energy management systems work better and save energy without sacrificing tenant comfort. Occupancy data is obtained by various methods, including high-precision instruments like thermal and optical cameras and specialized or environmental sensors like carbon dioxide (CO2) and passive infrared (PIR). While less accurate, the latter methods have drawn much interest since they are less expensive and invasive. This study's primary goal is to provide a unique prediction model that enhances the knowledge gleaned from sensor data by fusing evolutionary algorithms with machine learning (ML) models. To detect occupancy in an office setting, this study examines the effectiveness of four optimization algorithms: Multi-Verse Optimization (MVO), Teaching-Learning-Based Optimization (TLBO), Wind-Driven Optimization (WDO), and Whale Optimization Algorithm (WOA) when combined with Multilayer Perceptron (MLP). Among the noteworthy results is the strong performance of WDO-MLP, which shows remarkable AUC scores with little sensitivity to changes in swarm size. Furthermore, the best-performing approach is MVO combined with MLP, which shows better stability and discriminating ability at different swarm sizes. The practical implications of the research findings for implementing occupancy detection systems in smart buildings are noteworthy.
Dr. Halil Gör,
Volume 1, Issue 1 (9-2025)
Abstract
The accurate energy consumption prediction for OPEC (Organization of the Petroleum Exporting Countries) member states is vital for strategic planning and policy-making in the global energy market. This study leverages advanced machine learning techniques to forecast energy consumption, utilizing historical data from the U.S. Energy Information Administration (EIA). We applied a variety of machine learning models, including Simple Linear Regression, Gaussian Processes, Multilayer Perceptron (MLP), SMOreg, IBK, Kstar, LWL, Random Subspace, Random Committee, and Random Forest, to the task of predicting energy consumption. The performance of these models was evaluated based on metrics such as R-squared (R²), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Relative Absolute Error (RAE), and Root Relative Squared Error (RRSE). Our results demonstrated that the Random Committee model achieved the highest accuracy in both training (R² = 0.9999, MAE = 0.7411, RMSE = 1.0509, RAE = 1.2199%, RRSE = 1.2399%) and testing phases (R² = 0.9525, MAE = 11.4795, RMSE = 30.6585, RAE = 17.9586%, RRSE = 31.6700%), highlighting its robustness and predictive power. In contrast, the LWL model showed the poorest performance, with significant errors in both phases. The study also highlights the strengths and limitations of each model, with a focus on the applicability of these findings for policymakers and energy analysts. The insights gained from this research underscore the potential of machine learning to enhance energy consumption forecasting, providing a foundation for future studies to build upon. Directions for future research include incorporating additional socio-economic and environmental variables, real-time data, and more advanced machine learning techniques to improve prediction accuracy and reliability further.
Sarah A. Alabbas,
Volume 1, Issue 1 (9-2025)
Abstract
Given the huge growth in urban innovation over the past 10 years, smart cities require rational and workable solutions for transportation, building infrastructure, environmental conditions, and human enjoyment. This paper presents and explores data-mining-based predicted energy consumption models for a small-scale, intelligent steel company in South Korea. Devices built on the Internet of Things (IoT) are used to collect and process energy use data in order to forecast. Among the data used are leading and following currents, carbon dioxide emissions, load kinds, reactive power, and power factor. The Leagues Championship Algorithm (LCA), Evaporation-rate Water Cycle Algorithm (ERWCA), Multiverse Optimization Algorithm (MVO), Cuckoo Optimization Algorithm (COA), and Stochastic Fractal Search (SFS). The following metrics are used to evaluate the models' predictive power: root mean square error, mean absolute error (MAE), and coefficient of variation (R2) (RMSE). With the greatest R2 values (0.99800 during testing and 0.99815 during training), the Multilayer Perceptron (MLP) arrangement augmented by ERWCA performs exceptionally well and demonstrates a strong correlation between the predicted and actual energy consumption. Moreover, ERWCA has the lowest RMSE, a measure of the least number of prediction errors (2.09627 in the testing phase and 2.03778 in the training phase). Comparably low RMSE values (2.59167 during testing and 2.50700 during training) and excellent performance in terms of R2 values (0.99695 during testing and 0.99720 during training) also suggest that SFS could be used to improve MLP models for accurate energy consumption forecasts in smart city industrial buildings. These findings show how reliable and accurate the SFS and ERWCA energy use estimations are.
Samuel Nketia Boateng, Adwoa Tiwaah Ofori Atakorah,
Volume 1, Issue 1 (9-2025)
Abstract
It has been debated that to enhance data interpretation, and improve knowledge of environmental phenomena, Artificial Intelligence (AI) and Mixed Method Research (MMR) implementation and integration is of great value. However, not much is known about its actual use with environmental science data. Hence, the main aim of this study was to demonstrate how MMR and AI can improve environmental science data interpretation and analysis. The study adopted a case study titled “Willingness to Accept and Use Biogas Generated from Animal Manure and Agricultural Residue among University Students in Ghana” as the research that generated the data. A sequential explanatory MMR design was adopted. The survey collected data from (N= 231) and 5 in-depth interviews from Ghanaian University students. Through this study, it was found that majority of university students 112 (48.3%) are willing to install biogas in their future and homes and also 101(43.5%) are currently willing to use the energy. Again, the results showed that to integrate AI for a better understanding of environmental science research, researchers must first have a solid understanding of how to conduct MMR and obtain a reasonable picture of the main findings of the research conducted using statistical tools. Furthermore, the study found that AI was able to establish a relationship between both qualitative and quantitative data in an innovative way that provided answers to environmental issues, based on the results of the case study that was used. Additionally, the researcher must endeavor to supply appropriate prompts, the dataset, provide the framework that will guide the AI for enhanced data interpretation
