DEVELOPMENT OF A DEEP LEARNING NEURAL NETWORK MODEL FOR TRANSIENT AND SMALL SIGNAL STABILITY ASSESSMENT

ONYEMENAM,, JOHN OBIAJULU (2022) DEVELOPMENT OF A DEEP LEARNING NEURAL NETWORK MODEL FOR TRANSIENT AND SMALL SIGNAL STABILITY ASSESSMENT. Masters thesis, Landmark University, Omu Aran, Kwara State.

Text JOHN JUNIOR THESIS FINAL.docx Download (988kB)

Abstract

This study's objective is to assess the degree of instability that existing power systems are subject to as a result of incorporating novel elements such power electronics, electric vehicles, and renewable energy production. The development of renewable energy sources is currently having an impact on the reliability and security of the electrical network. Due to the potential for disastrous power outages, a wide range of stakeholders are paying attention to transient stability and tiny signal stability issues. With the help of a feature selection and DLNN technique, the aim of this research is to evaluate the numerous stability issues relating to the electricity system. Data contingencies for the Nigerian 28 bus system and IEEE 9 bus system were produced using DIgSILENT. When the line on DIgSILENT was removed, the fault was applied and cleared. A data processing pipeline for feature selection is built using the Relief-F feature selection approach. A DLNN model was created in the Python environment to train and anticipate the power system's transient stability and tiny signal stability. The prediction model will advise the power system operator on how low frequency local and inter-area oscillations will be suppressed while a system is transiently stable. In order to implement the necessary control mechanisms, the DLNN approach also gives information on the system's transient stability and oscillatory dynamic response. The DIgSILENT/Python application, which utilizes an Intel Pentium core i5 2GHz CPU, is used to carry out this analysis. The Nigeria 28 bus system and the IEEE 9 bus system are used to test the increased performance of the suggested model. The results demonstrate evaluation performance metrics for accuracy, precision, sensitivity, f1-score, specificity, mean squared error, and root mean square error for the Nigeria 28 bus system and the IEEE 9 bus system. The Nigeria 28 bus system and the IEEE 9 bus system's evaluation metrics were compared to other works in the related literature. This study shows how the DLNN technique can be used to evaluate transient stability and tiny signal stability in real-time, online.

Actions (login required)

View Item