Neural-Network Load Forecasting and PV-Supported Planning of the Bab Al-Azizia 30-kV Network for 2026-2031

Abstract

This research paper proposes a neural network-based method using MATLAB to predict the annual peak load for the 30 kV electricity distribution network in the Bab Al-Aziziya loop from 2026 to 2031, along with the use of photovoltaic energy to enhance network efficiency. The study data shows an annual load growth rate requiring compensation to keep the network operational until 2030, followed by operational disruptions in 2031 affecting the 220/30 kV main transformer, the 30/11 kV backup transformer, and the 30 kV Al-Farnaj distribution bus. A feedforward neural network was developed using MATLAB to predict the annual peak load, and the prediction accuracy was evaluated using the mean absolute error (MAE), root mean square error (RMSE), and mean percentage absolute error (MAPE). The peak electricity load is projected to increase from 126,276 MW in 2026 to 185,541 MW in 2031, underscoring the need to strengthen the electricity grid to meet future demand growth. To mitigate operational constraints, a support plan for photovoltaic power generation is being evaluated using current energy flow calculations. Before the photovoltaic system was connected to the grid, energy consumption was 185,541 MW; after connection, energy consumption decreased to 148,370 MW, equivalent to 37,171 MW of subsidized electricity from the photovoltaic system, representing a 20.03% reduction in energy consumption. Furthermore, the minimum distribution bus voltage increased from 94.89% to 97.35%, eliminating the overvoltage issue observed in the distribution bus in the Furnaj area. The results show that combining neural network-based load prediction with strategically placed photovoltaic power generation can provide an effective planning tool to improve voltage characteristics, reduce network load, and enhance the operational flexibility of medium-voltage distribution networks.