Forecasting Photovoltaic Power Production and Energy Consumption Using Artificial Neural Networks: Case Study of a Residential Microgrid in Morocco

This chapter delves into the critical role of accurate energy forecasting in optimizing the use of renewable energy sources, with a focus on photovoltaic (PV) power production and energy consumption. The study utilizes a real energy dataset collected from residential living labs in Benguerir, Morocco, to implement and compare two advanced forecasting methods: Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) networks. The chapter provides a detailed overview of neural network methods for energy forecasting, highlighting the architecture and learning processes of MLP and LSTM. The energy forecast methodology is thoroughly described, including data preparation, model building, and evaluation analysis. The case study presents a comprehensive analysis of PV production and energy demand patterns, showcasing the forecasting results and statistical measurements that assess the performance of both models. The findings reveal that while both MLP and LSTM demonstrate high-quality performance, the MLP model offers more accurate results and better captures the variance in the data. This chapter offers a deep dive into the practical application of neural networks in energy forecasting, providing valuable insights for those seeking to enhance the efficiency and reliability of renewable energy systems.