Search

Accurate estimation of wind speed distributions is a challenging task in wind power planning and operation. The selection of convenient functions for describing wind speed distribution is a crucial requisite. In this paper, remarkable bi-parameter Weibull function is presented to estimate the wind energy potential. Weibull parameters based on different six estimation methods, namely graphical, method of moment, energy pattern factor, mean standard deviation, power density methods, and genetic algorithm are evaluated. Besides, the goodness of fit of the estimation methods is investigated via mean absolute error, root mean square error, normalized mean absolute error, Chi-square error, and regression coefficient. To plainly identify the best matching estimation method, Net Fitness test is also presented. Catalca in the Marmara region in Istanbul, Republic of Turkey, is selected to be the underlying site. The experimental results show the effectiveness of the estimation methods in modeling wind distribution but with relatively small differences in terms of performance. However, the genetic algorithm and energy pattern factor accomplish the best and worst matching estimation methods, respectively. [ABSTRACT FROM AUTHOR]

Photovoltaic (PV) panels are used to generate electricity by using solar energy from the sun. Although the technical features of the PV panel affect energy production, the weather plays the leading influential role. In this study, taking into account the power of the PV panels, the solar energy value it produces and the weather-related features, day-ahead solar photovoltaic energy forecasting is carried out over three different long short-term memory (LSTM) networks: LSTM, bidirectional long short-term memory (BiLSTM) and stacked LSTM. Finally, a comparative study with LSTM, BiLSTM and stacked LSTM models is constructed by using an actual dataset obtained with a 1741-day-long for 26 different panels on an inverter in İstanbul, Turkey. Stacked LSTM model for PV power forecasting predictions show that the average performance metrics root mean squared error of the 26 datasets reach to 19.17 kWe and 18.92 kWe for 50 and 100 epochs, respectively. The findings suggest that the proposed model is a reliable technique for solar photovoltaic energy prediction. [ABSTRACT FROM AUTHOR]