A Survey of AI-Powered Mini-Grid Solutions for a Sustainable Future in Rural Communities

TL;DR

This paper surveys AI-powered mini-grid solutions aimed at improving sustainable energy access in rural communities, focusing on forecasting the inherently variable generation from renewables and the demand of rural loads. It categorizes forecasting techniques into physical prediction models, statistical models, intelligent computational methods, and hybrids, and discusses evaluation metrics such as $MAE$, $MSE$, $RMSE$, $MAPE$, and $SMAPE$ across very short- to long-term horizons. It reviews public datasets and tools including Prophet, NeuralProphet, and N-BEATS, and analyzes energy-generation, energy-demand, and energy-management forecasting for mini-grids, highlighting the role of multi-objective optimization in battery scheduling. The paper concludes with recommendations on model adaptation, transfer learning, and horizon selection to address real-world data delays and local cultural factors, aiming to inform practical deployment of AI-driven mini-grid solutions.

Abstract

This paper presents a comprehensive survey of AI-driven mini-grid solutions aimed at enhancing sustainable energy access. It emphasises the potential of mini-grids, which can operate independently or in conjunction with national power grids, to provide reliable and affordable electricity to remote communities. Given the inherent unpredictability of renewable energy sources such as solar and wind, the necessity for accurate energy forecasting and management is discussed, highlighting the role of advanced AI techniques in forecasting energy supply and demand, optimising grid operations, and ensuring sustainable energy distribution. This paper reviews various forecasting models, including statistical methods, machine learning algorithms, and hybrid approaches, evaluating their effectiveness for both short-term and long-term predictions. Additionally, it explores public datasets and tools such as Prophet, NeuralProphet, and N-BEATS for model implementation and validation. The survey concludes with recommendations for future research, addressing challenges in model adaptation and optimisation for real-world applications.

Figures (9)

• Figure 1: An example of how the temperature variable can be defined by a fuzzy set of category memberships
• Figure 2: Left: RBF kernel with non-linear boundary. Right: Linear kernel with linear boundary.
• Figure 3: An example of the structure of a simple neural network featuring 2 hidden layers
• Figure 4: Illustration of an RNN being unfolded across three time steps, where the recurrent connections are expanded to show the flow of information from input ($x$) to hidden state ($h$) and output ($o$) at each time-step, with weight matrices $U$, $W$ and $Z$ shared across all steps rnn.
• Figure 5: An example of an LSTM cell featuring forget, input, and output gates