Monthly Electricity Production Forecast with Deep Learning in Solar Power Plants

The total installed power in Turkey is 97,070 MW, of which renewable energy sources provide 64%. Solar power plants constitute 7.2% of renewable energy sources. It is desired to reduce the use of fossil fuels in electricity generation and to meet the increasing electricity demand in our country with renewable energy. In this context, the demand for solar power plants is increasing day by day. However, one of the critical factors that worry investors is the amortization period. A study was conducted on estimating this time using artificial intelligence and deep learning applications. In this thesis, LSTM (Long Short Term Memory) artificial neural networks were used. With machine learning, it has been estimated how much electricity the solar power plant can produce in the coming months. The calculation used parameters such as weather conditions, solar panels, inverters, and performance ratio of the past years. As a result of the analyzes made, an error rate between 1% and 17% was encountered. It has been seen that this rate is much more successful than other artificial intelligence algorithm predictions.