Application of Artificial Neural Networks in Solar Photovoltaic Power Forecasting
Data-Driven Energy Management and Tariff Optimization in Power Systems: Shaping the Future of Electricity Distribution through Analytics, wiley, ss.167-177, 2025
- Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
- Basım Tarihi: 2025
- Doi Numarası: 10.1002/9781394290307.ch8
- Yayınevi: wiley
- Sayfa Sayıları: ss.167-177
- Anahtar Kelimeler: Artificial neural networks (ANNs), mean squared error (MSE), photovoltaic (PV) panels, power forecasting
- Atatürk Üniversitesi Adresli: Hayır
Özet
Huge value of energy demands, limited resources of fossil fuels, global warming, and climate changes have caused an urgent need for renewable energy sources (RES), energy storages, and management strategies. Meanwhile, penetration of RES in energy systems has been faced with some challenges, such as very short-term fluctuations of tidal, wind, and solar products. Hence, installation of energy storage in RES-oriented systems can mitigate this crisis significantly. However, uncertain parameters such as demand, energy market price, RES outputs, and weather conditions impose some systematic, economic, and environmental considerations on optimal planning, operation, and maintenance scheduling of storage-coupled facilities. Artificial neural networks (ANNs) are able to forecast RES products in water, gas, and power systems accurately. Therefore, this chapter aims to present a novel algorithm for short-term forecasting of photovoltaic power using ANNs. Moreover, mean squared error (MSE) of estimated values is calculated and minimized as an objective function under the Python programming language. Simulations are conducted on a real solar field located in Ahwaz, Iran, to predict its hourly power product during one week.