نرجس عزيزي

The intermittence and fluctuation character of solar irradiance causes severe limitations on most of its applications. The precise forecast of solar irradiance is the critical factor in predicting the output power of a photovoltaic power generation system. To this end, in the present study, the feasibility of a solar photovoltaic power plant in Sistan and Baluchistan province has been investigated with the help of machine learning methods. In this regard, six cities have been selected according to their geographical location. Then, solar radiation was forecasted for the future ten years in each city based on machine learning methods, including MLP, LSTM, GRU, CNN, and CNN-LSTM. The input parameters consist of single and multivariate features, and the output results are extracted directly and recursively. For each city, the best prediction model was selected according to the eva‎luation parameters such as RMSE, R2, MAE, and NRMSE, helping to compare the radiation output results with each other. According to historical and forecasted data, the city of Zahedan was selected to have the best global horizontal radiation. The amount of power output from the 20 MW power plant in Zahedan city during the years 2022 to 2031 was obtained on an average of 50.0184 GWh per year. From the economic point of view, the time of return on investment occurs nearly 12 years after the construction of the power plant, and by giving government incentives, reducing the costs of installing panels and raising the price of purchasing electricity from power plants, the time of return on investment can be shortened. improve the The leveled cost of electricity in this city was 23.82 dollars per megawatt hour.

يادگيري ماشين , تابش افقي جهاني , پتانسيل سنجي , شبكه عصبي مصنوعي , شبكه عصبي بازگشتي , سيستان و بلوچستان

Machine Learning , Global Horizontal Radiation , Potential Measurement , artificial neural network , Recurrent Neural Network , Sistan and Baluchistan

Narjes Azizi

Meisam Farajollahi