"Integrating AI and AEM Electrolyzer for Green Hydrogen Production: Optimization of Solar-Powered Electrolysis in Residential Energy Management"

This master's thesis presents a comprehensive study on the forecasting of short-term power generation in a grid-connected hybrid solar photovoltaic (PV) system through the utilization of an artificial intelligence (AI) model. The research integrates weather data and solar PV electricity production data to develop and optimize a Long Short-Term Memory (LSTM) based AI model. The year 2021's solar PV and weather data were utilized for training and validating the model. Additionally, AEM electrolyzer was optimized to efficiently produce hydrogen using surplus electricity generated by the solar PV system . The investigation identified notable correlations between solar radiation, solar energy, UV index, and various other weather parameters with solar PV power generation. These correlations played a significant role in enhancing the accuracy of the AI model in predicting power generation. Various LSTM model structures were evaluated, and a two-layer LSTM model demonstrated superior performance, achieving an accuracy of approximately 80%. Furthermore, surplus electricity generated by the system, averaging 10 kWh during the daytime was calculated and analyzed. The economic viability of the hybrid system was also established, as the cost of electricity generated through the hybrid system was less than half of the grid energy price, meeting the regulatory standards.Optimizing the AEM electrolyzer revealed that a configuration with a few standby parallel AEM electrolyzers was optimal for utilizing excess electricity effectively. Further than that scheduling the parallel system in hourly basis for the days ahead, would help to have more conveniently benefit from this system. In conclusion, this research presents promising avenues for future studies aimed at further enhancing the efficiency and sustainability of renewable energy systems. Prospective research includes real-time integration of weather updates for AI models, advanced energy storage systems, demand-side management strategies, comparison of machine learning algorithms, optimized hydrogen production, and the evaluation of the integrated model in a microgrid setting. These future directions aim to contribute to the wider adoption of renewable energy sources and facilitate the transition towards a more sustainable energy future.