On Sunday 17-11-2024, the discussion of the master's thesis of the student

Saja Jaafar Jawad, entitled

Electricity Outage Estimation in Iraq

The committee included:

1- Prof. Dr. Hanan Abdul-Ridha Akkar - University of Technology - Department of Electrical Engineering - Chairman

2- Asst. Prof. Dr. Shaima Safaa Al-Din Baha Al-Din - University of Nahrain - Department of Computer Engineering - Member

3- Asst. Dr. Ahmed Hani Youssef - University of Nahrain - Department of Computer Engineering - Member

4- Asst. Dr. Shaima Walid Abdul Latif - University of Nahrain - Department of Computer Engineering - Member and Supervisor

The student fulfilled the requirements for obtaining a master's degree in computer engineering

The thesis included the following:

Power outages in Iraq have been a widespread and ongoing problem for decades, disrupting the economy, daily life, and essential services such as hospitals and government offices. This necessitates finding fundamental solutions, and predicting outages that can help authorities implementing proactive measures to understand and mitigate the number of outage hours. This thesis uses temperature and electrical load time series data to predict and identify various electricity outage patterns. Long-term and short-term outages were forecasted for Baghdad, and the impact of these variables on outages was analyzed. For accurate outage prediction, advanced deep learning techniques capable of handling time series, such as LSTM, RNN, GRU, and 1D CNN, were employed, combined with various optimization methods like MRFO, PSO, Random Search, and Bayesian for hyperparameter selection to ensure the effectiveness of the deep learning techniques. Performance was evaluated using error metrics such as MSE, RMSE, and MAE. The results of short-term outage prediction show that the combined MRFO with GRU outperforms the other models used in this research in terms of accuracy with MSE (0.000234) and RMSE (0.015300). Regarding processing time, the combined Bayesian with CNN outperforms the other models, taking two minutes and twenty-seven seconds. As for the long-term prediction results, the combined MRFO with LSTM achieves the best results with MSE (0.00522) and RMSE (0.072259), but in terms of processing time, the combined Bayesian with LSTM is the best and takes two minutes and one second. Python programming language was used to develop the proposed models to predict power outages in Baghdad