
Using Innovative AI Techniques For STLF, Under Limited Data Availability
Abstract:
This thesis addresses the critical need for accurate Short-Term Load Forecasting (STLF) in Pakistan's energy sector, where traditional methods fail to meet the challenges posed by a decade-long energy shortage, load shedding, and unreliable consumption data. With a renewable energy mix, the power system faces new uncertainties that demand sophisticated forecasting techniques to ensure stability and efficiency. The research presents three key contributions tailored to these challenges. First, a distributive STLF algorithm is introduced, designed to predict electricity usage using only recent load patterns without relying on external data like weather or socio-economic factors. This algorithm is particularly suitable for large-scale applications, thanks to its scalability and simplicity, making it highly adaptable to Pakistan’s diverse consumer base. The second contribution is a multi-step STLF method that uses minimal data inputs—specifically electricity usage, weather, and calendar data—yet can be applied at varying levels of temporal and spatial granularity. This flexibility allows for precise forecasting from individual households to larger consumer groups, facilitating the integration of renewable resources and enabling more dynamic and responsive grid management. Finally, the third work integrates spatial and temporal data through a novel Spatio-Temporal STLF model. Unlike traditional methods that only use temporal data, this approach harnesses the power grid's graphical structure to capture spatial correlations between consumers. By extracting spatial dependencies and feeding them into neural networks, the model enhances forecasting accuracy, providing a more holistic view of load patterns. These innovative algorithms offer significant advancements in managing Pakistan’s energy challenges. By providing more accurate and adaptable load forecasts, they aim to improve grid stability, optimize energy distribution, and support the seamless integration of renewable energy sources, paving the way for a more sustainable and resilient energy future for the country.
Thesis Defense Committee:
Dr. Mian Muhammad Awais (Professor, Thesis Committee Member), Department of Computer Sciences, Lahore University of Management Sciences (LUMS), Pakistan
Dr. Asim Karim (Professor, Thesis Committee Member), Department of Computer Sciences, Lahore University of Management Sciences (LUMS), Pakistan
Dr. Basit Shafiq (Associate Professor, PhD Supervisor), Department of Computer Sciences, Lahore University of Management Sciences (LUMS), Pakistan
Dr. Nauman Zafar Butt (Associate Professor, Thesis Committee Member), Department of Electrical Engineering, Lahore University of Management Sciences (LUMS), Pakistan
Dr. Awais Yasin (Associate Professor, External Examiner), Department of Computer Engineering, NUTECH University, Pakistan
Dr. Naveed Arshad (Associate Professor, PhD Supervisor), Department of Computer Sciences, Lahore University of Management Sciences (LUMS), Pakistan