Short-Term Wind Speed Prediction Base on Backpropagation Levenberg-Marquardt Algorithm; Case Study Area Nganjuk

Wind energy is one of the main sources of renewable energy that is widely converted to electrical energy because reduce emission. The fluctuating of the wind will affect power quality of wind power generation when delivered to grid so that it takes prediction of short-term wind speed. In this study, wind speed prediction use Backpropagation Neural Network (BPNN) with Lavenberg - Marquardt algorithms for weight update. Its performance will be compared with Conjugate Gradient Fletcher Reeves (CGFR) and Conjugate Gradient Quasi Newton (BFGS) based on correlation coefficient, Mean Square Error (MSE), Mean Absolute Percentage Error (MAPE) and computation time. Based on the testing, Levenberg Marquardt Backpropagation Neural Network is the most optimal algorithm compared to fletcher - reeves and quasi newton where the value of correlation coefficient is 0.99056, MSE 0.0187, MAPE 5.1965 and computation time 79.19 milisecon.