Optimal OLTC Control Using ML Algorithms for Synchronous Generator Integration with Off-Grid Renewable Energy in Kenya

Optimal voltage control is desirable in an electrical power system for voltage stability. The Kenyan energy sector has experienced growth in isolated renewable energy grids in the wake of the net global shift to renewable energy sources (RES) projected to slow down the carbon footprint. Solar photovoltaic and wind energy are the most dominant RES which cope with grid reactive power requirements by relying on switched power-electronic devices. The mismatch in synchronizing speed causes these devices to often fall out of step with synchronous generators (SyG) whenever a system disturbance occurs. Voltage and frequency instability can be mitigated by the integration of synchronous generators with load-following capability and reactive power compensation capability as proposed by previous studies. However, supplementary voltage regulation through the main transformer's on-load tap changer (OLTC) is required. This paper proposes the use of artificial intelligence (AI) for OLTC control and compares the applicable advanced machine learning (ML) algorithms for the power transformer OLTC control in order to recommend the most accurate method. AI leverages on multiple parameters in the prediction of the appropriate percentage tap position which is not conventional to many digital OLTC relays available in the Kenyan grid.