This blog post wraps up my graduate research journey, encompassing my Master's studies at Southern Illinois University Carbondale (SIUC) from 2011 to 2012 under the supervision of Prof. Constantine J. Hatziadoniu, and my Doctoral research at the University of North Carolina at Charlotte (UNC Charlotte) from 2014 to 2018 under the supervision of Prof. Badrul Chowdhury.
The rapid expansion of variable renewable energy (VRE) sources, such as wind and solar photovoltaic (PV) systems, introduces significant uncertainty and operational challenges into modern electric power grids. To support high-penetration renewable integration and enhance grid situational awareness, my graduate research focused on developing data-driven predictive, descriptive, and prescriptive analytics alongside advanced heuristic optimization. The following sections summarise my research activities organized chronologically by core thematic fields, including related publications, master's and doctoral theses, and their official links.
Theme 1: Optimization & Sensitivity Methods for Wind and Integrated Power Systems
- For the foundational analysis regarding particle swarm optimization layouts applied to baseline power networks embedded with wind generators, please refer to my Master's Thesis: Study of particle swarm for optimal power flow in IEEE benchmark systems including wind power generators (SIUC Thesis).
- For exploring sensitivity analysis approaches to select the most effective control variables when solving optimal power flow problems using particle swarm optimization, please refer to our arXiv preprint on control variables selection.
- For details on the evaluation of the economic dispatch model for the seamless integration and optimal allocation of wind power generation using particle swarm optimisation, please refer to our arXiv preprint on integrated wind systems economic dispatch.
Theme 2: Machine Learning & Ensemble Architectures for Solar Power Point Forecasting
- For analyzing the application of Artificial Neural Networks (ANN) in establishing non-linear curve-fitting mapping structures for solar generation forecasting, please refer to our IEEE NAPS 2015 conference proceeding on ANN-based forecasting.
- For a deep assessment of Support Vector Regression configurations and baseline comparative models in deterministic solar forecasting, please refer to our ASEM 2016 conference proceeding on Support Vector Regression modeling.
- For details on using a Random Forest to ensemble multiple SVR models for robust day-ahead predictions, please refer to our IEEE ISGT 2017 conference proceeding on Random Forest Ensembles of SVRs.
Theme 3: Probabilistic Forecasting & Uncertainty Quantification
- For exploring simple linear statistical setups to formulate a baseline probabilistic forecasting framework for variable solar generation outputs, please refer to our IEEE SoutheastCon 2015 publication on Multiple Linear Regression analysis.
- For utilizing Random Forest ensemble learning structures to generate hour-ahead nonparametric probabilistic forecasts that measure baseline risks and uncertainties, please refer to our IEEE NAPS 2017 publication on Hourly Probabilistic Forecasting.
Theme 4: Solar Power Ramp Events Forecasting and Post-Processing
- For the overarching framework, methodologies, and full scope of our data-driven predictive mitigation strategies, please refer to my full PhD Dissertation: A Post-Processing Approach for Solar Power Combined Forecasts of Ramp Events (UNC Charlotte Dissertation).
- For the development of a data-driven post-processing adjusting approach targeted at resolving high-error forecast anomalies and improving hour-ahead combined forecasts during sharp ramp events, please refer to our Renewable Energy (Elsevier) journal paper.
- For incorporating estimated ramp rates to mitigate systemic forecasting errors and optimize final blended solar generation forecasts, please refer to our IET Renewable Power Generation journal paper.
- For framing solar power ramp detection as a categorical problem through advanced machine learning classification techniques, please refer to our IEEE PEDG 2018 conference proceeding.
- For assessing the descriptive quality and validation of combined forecasting methods from a ramp-event perspective, please refer to our IEEE PES General Meeting 2018 conference proceeding.
- For evaluating the implementation of our data-driven adjusting post-processing scheme specifically applied to very short-term solar PV power predictions, please refer to our IEEE Maghreb Meeting 2021 conference proceeding.