ANN-Based Peak Load Prediction and MMC-Solar PV with ANFIS MPPT

This book presents an integrated approach that combines peak load prediction and advanced control of a grid-connected solar PV system to support distribution substations. Peak load forecasting for the Bahir Dar 15 kV substation is carried out using an Artificial Neural Network trained with newly registered domestic customers, their sanction loads, and corresponding substation peak load data. The model predicts that the substation will face overloading within 22 months. Because the peak demand occurs during daytime and the city has strong solar resources, a solar PV system is proposed as a peak load shaving solution and as an ancillary support source.A 100 kW PV system is designed to replace the diesel generators at the EEU data center. An ANFIS-based boost converter regulates the DC-link voltage under changing environmental conditions. A five-level Modular Multilevel Converter interfaces the PV system with the grid, using voltage-oriented control, circulating current suppression, and capacitor voltage balancing. The designed system achieves low harmonic distortion, with injected current THD reduced to 1.12%, well below the 5% international standard.