An innovative AI-based forecasting model developed by Halima Haque, a student of Electrical and Electronic Engineering (EEE) at Independent University, Bangladesh (IUB), aims to enhance Dhaka’s power supply system and minimize load shedding.
The model can predict electricity demand across residential, commercial, and industrial sectors, helping to ensure efficient power distribution, especially during peak summer months. This advancement is expected to reduce load shedding and promote optimal use of renewable energy.
Supervised by Dr. Md. Abdur Razzak, Professor at IUB, the research analyzed over 6.5 million data points collected from Dhaka Electric Supply Company Limited (DESCO) between 2020 and 2023. The study considered factors like weather patterns, consumption behaviors, and tariff structures to build a precise forecasting model. 
“To meet the growing demand for electricity due to rapid urbanization and industrialization, accurate demand forecasting is essential,” said Halima. “This AI-powered technology will assist power distribution companies and policymakers in making informed decisions, leading to a more stable and sustainable power supply.”
The research tested five machine learning models — K-Nearest Neighbors, Random Forest, Light Gradient Boosting Model, Extreme Gradient Boosting, and Multiple Linear Regression. The findings were presented at IEEE conferences in the Maldives and Thailand and published in the prestigious IEEE Transactions on Industry Applications journal.
“This is a crucial step towards modernizing Bangladesh’s power infrastructure,” said Dr. Razzak. “By leveraging machine learning, we can better understand electricity usage patterns, create more effective policies, and pave the way for sustainable urban development.”
The AI model’s success marks a significant milestone in using advanced technology to address real-world energy challenges, positioning Bangladesh as a leader in smart grid innovation.
