Projects

Project Title: Controller Design for DC Grid-Following/Forming Converter Systems [Summer 2025 - Present]

📌 Developed control theorem for DC grid-following/forming converter systems through MATLAB Simulink.

📌 Validated the lab results through an experimental testbed.

💻 Manuscript is under preparation

Project Title: Nested Controller Design for Scaled Up-/Down DC/DC Converter Systems [Spring 2025 - Summer 2025]

📌 Designed a nested control theorem for DC/DC buck and boost converter systems to reduce high-scale power system building costs.

📌 Evaluated software simulation outcomes through lab hardware experiments.

💻 Manuscript is under preparation

Project Title: Controller Design for High and Low Voltage DC/DC Converter Systems [Fall 2024 - Spring 2025]

📌 Developed a single-loop control theorem for testing high and low voltage DC/DC buck converter systems in the laboratory before field implementation.

📌 Compared the results through both software and hardware experiments.

💻 Manuscript is under review

Project Title: Techno-Economic Analysis of Second-Use EV Batteries [Spring 2024 - Spring 2025]

📌 Conducted a techno-economic analysis of retired EV batteries with sufficient state-of-health (SOH) for DC fast charging applications to overcome the increase in power demand on the grid during peak hours.

📌 Developed an adaptive business model for EV charging stations focusing on retired batteries, to address the challenges raised, including economic inflation, compared to new batteries.

💻 Manuscript is under review

Project Title: Design Experimental Twins Between Low and High Voltage Inverters [Fall 2023 - Spring 2024]

📌 Designed a dynamic power controller to evaluate high-voltage inverter systems without the need to build expensive, large-scale inverter-based resources (IBRs) for testing.

📌 Validated the equivalence of the controller designed through electromagnetic simulation and hardware experiments.

Fig. Full inverter interfaced IBRs from [Article -AC Experimental Twin]

Fig. Illustration showing an overall integration of a grid-tied IBR system from [Article -AC Experimental Twin]

Fig. Simulation model of a grid-connected solar PV system from [Article -AC Experimental Twin]

Fig. EMT simulation comparison for actual vs. per-unit controllers of a 20kW IBR: a) dc-link voltage, b) d-axis current, c) PCC active power, d) PCC reactive power, from [Article -AC Experimental Twin]

Fig. EMT simulation comparison for actual vs. per-unit controllers of a 2MW IBR: a) dc-link voltage, b) d-axis current, c) PCC active power, d) PCC reactive power, from [Article -AC Experimental Twin]

Fig. EMT simulation comparison for actual vs. per-unit controllers of a 2kW IBR: a) dc-link voltage, b) d-axis current, c) PCC active power, d) PCC reactive power, from [Article -AC Experimental Twin]

Fig. Experimental setup from [Article -AC Experimental Twin]

Fig. Experimental comparison for actual (a1-a3) vs. per-unit (b1-b3) controllers of a 2kW laboratory IBR: (a1)/(b1) dc-link voltage, (a2)/(b2) d-axis current, c) (a3)/(b3) 3-phase PCC current, from [Article -AC Experimental Twin]

📍Article -AC Experimental Twin: S. Li, H. Mondal, M. R. Rafi, and Y. -K. Hong, "Building Experimental Twins Between Low and High Voltage Inverters Based on A Novel Per-Unit and Time-Angular Domain Conversion Technique," IEEE Transactions on Power Electronics, August 2025.

🔗 Article-AC Experimental Twin

Project Title: Artificial Intelligence for Building Power and Energy Management Systems [Fall 2021 - Summer 2023]

📌 Designed a priority-based deep learning architecture, i.e., Deep Weighted Fusion Learning (DWFL), to detect and predict building occupancy counts with an optimal multi-sensor fusion structure.

📌 Achieved a 3% improvement in performance compared to baseline models.

Fig. The overview of the multi-sensor learning and fusion system from [Article -DWFL]

Fig. Deep Weighted Fusion Learning model from [Article -DWFL]

Fig. HOBO MX logger from [Article -DWFL]

Fig. Occupancy detection from image data from [Article -DWFL]

Table. Summary of lab test plan from [Article -DWFL]

Table. Model performance analysis considering environmental and room size effects from [Article -DWFL]

Fig. Occupancy prediction performance analysis for different types of sensors from [Article -DWFL]

Table. Comparison of proposed DWFL method with baseline models from [Article -DWFL]

📍Article -DWFL: M. R. Rafi, F. Hu, S. Li, A. Song, X. Zhang, and Z. O’Neil, “Deep weighted fusion learning (DWFL)-based multi-sensor fusion model for accurate building occupancy detection”, Energy and AI, vol. 17, September 2024

🔗 Article-DWFL

Project Title: AI-based Cybersecurity [Spring 2021 - Summer 2021]

📌 Applied UNSW-NB15 dataset (available online) containing 210,000 data points, including 61,987 attack data and 148,013 benign data.

📌 Performed machine learning algorithms such as k-nearest neighbors (KNN), logistic regression, and random forest, resulting in 98.03% accuracy in detecting attacks on Internet of Things (IoT) networks.

Fig. K-nearest neighbors (KNN) model architecture from [Article -Cybersecurity]

Fig. ROC Curve for KNN from [Article -Cybersecurity]

Fig. Logistic Regression (LR) model architecture from [Article -Cybersecurity]

Fig. ROC Curve for LR from [Article -Cybersecurity]

Fig. Random Forest (RF) model architecture from [Article -Cybersecurity]

Fig. ROC Curve for RF from [Article -Cybersecurity]

📍Article -Cybersecurity: M. R. Rafi, M. A. Salam, and O. Kandara, “Machine Learning Based Attack Detection in Internet of Things Network,” International Journal of Computer Science and Information Security, vol. 19, no. 8, August 2021

🔗 Article-Cybersecurity

Project Title: Application of Machine Learning in Material Science [Spring 2020 - Summer 2021]

📌 Utilized Convolutional Neural Network (CNN) and Random Forest Regressor models for predicting the yield strength of high entropy alloys (HEAs) with lab experimental validation.

📌 Predicted yield strength of HEAs (MoNbTaTiW and HfMoNbTaTiZr) in the lab at desired high temperatures with a low, i.e., 2.5% and 6.4% error margin, respectively.

Fig. Heat-map of the correlation matrix between the features present in the datasets from [Article -Material Science]

Fig. Scatterplot matrices showing the relationships between all the features present in the 240 datasets from [Article -Material Science]

Fig. Schematic diagram of random forest regressor model showing multiple decision trees with majority voting from [Article -Material Science]

Fig. RF regressor model prediction of yield strength using RF regressor model for (a) MoNbTaTiW (b) HfMoNbTaTiZr. The red point represents the yield strength prediction of HEAs at 800 ◦C. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the [Article -Material Science])

Table. The yield strength (MPa) prediction of MoNbTaTiW and HfMoNbTaTiZr at 800◦C, 1200 ◦C, and 1500 ◦C by using the RF regressor model from [Article -Material Science]

📍Article -Material Science: U. Bhandari, M. R. Rafi, C. Zhang, and S. Yang, “Yield strength prediction of high-entropy alloys using machine learning,” Materials Today Communications, vol. 26, March 2021

🔗 Article-Material Science

Project Title: Application of Machine Learning in Wireless Networks [Spring 2018 - Spring 2019]

📌 Generated transport blocks (TBs) before signaling using attributes for uplink transmission, based on a random forest prediction algorithm, to overcome the latency in Long Term Evolution (LTE) networks.

📌 Reduced latency in the LTE network by 2 and 8 milliseconds for a transport block during uplink data transfer and uplink or downlink data retransmissions, respectively.

Fig. State-of-the-Art Uplink Data Transfer with Hybrid Automatic Repeat Request (HARQ) from [Article -Latency]

Fig. Proposed Uplink Data Transfer with HARQ from [Article -Latency]

Fig. State-of-the-Art Downlink Data Transfer with HARQ from [Article -Latency]

Fig. Proposed Downlink Data Transfer with HARQ from [Article -Latency]

📍Article -Latency: M. T. Kawser, M. R. Rafi, F. M. Faijus Salehin Rifat, M. F. Rahman, N. Zaki, and S. Z. Rahman, “Prior Generation of Transport Blocks to Expedite HARQ Retransmissions and Uplink Transmissions Improving Latency in LTE,” 2019 IEEE 5th International Conference on Computer and Communications (ICCC), 2020

🔗 Article-Latency

Page updated

Google Sites

Report abuse