International Journal of Power Electronics and Drive Systems (IJPEDS) Vol. 14, No. 3, September 2023, pp. 1345~1355 ISSN: 2088-8694, DOI: 10.11591/ijpeds.v14.i3.pp1345-1355  1345 Journal homepage: http://ijpeds.iaescore.com Prediction model of wind speed using hybrid artificial neural network based on Levenberg-Marquardt algorithm Anas Elmejdki 1 , Khalid Hati 2 , Hilal Essaouini 1 1 Energy Laboratory, Department of Physics, Faculty of Sciences, Abdelmalek Essaadi University, Tetouan, Morocco 2 Systems of Communications and Detection Laboratory, Department of Physics, Faculty of Sciences, Abdelmalek Essaadi University Tetouan, Morocco Article Info ABSTRACT Article history: Received Jan 13, 2023 Revised Mar 8, 2023 Accepted Mar 29, 2023 In this paper, a new method is developed to model the wind speed data that is considered as a function of seasonal wind variations. A hybrid artificial neural network (HANN) is investigated based on the Weibull distribution model. The presented HANN model predicts wind speed data with seasonal and chronological characteristics similar to real wind data. The design of the wind farm was implemented using MATLAB software. The suggested model has been applied and validated with wind data collected from the site of Tangier-MED in Morocco over two years, 2015 and 2016. The errors in terms of mean absolute percentage error MAPE and root mean square error RMSE are respectively 0.011 and 0.067 in 2015. Keywords: Artificial neural network Levenberg-Marquardt Optimization Weibull distribution Wind energy Wind speed This is an open access article under the CC BY-SA license. Corresponding Author: Anas Elmejdki Energy Laboratory, Department of Physics, Faculty of Sciences, Abdelmalek Essaadi University Tetouan, Morocco Email: amejdki@gmail.com 1. INTRODUCTION In the last two decades, the world has been increasingly turning to sustainable energy sources [1] such as solar [2]–[4] wind [5], [6] and waves to benefit from and to use in order to replace conventional fossil fuels. However, these sources are insufficient to meet the demands of the ever-growing population [7]. Unlike conventional generators, renewable energy generators can only produce energy when green energy resources are available. As a result, accurate prediction, control, and representation of renewable energy systems are critical [8] to ensuring a stable and continuous energy supply. The properly representation of probabilities, uncertainties, and fluctuating behaviors of the renewable energy systems allow them to be accurately optimized [9]. Wind energy is random and volatile, and its speed is the main parameter that influences wind power. Accurate wind speed prediction is beneficial to power system operation, security analysis, peak regulation, and energy saving. Therefore, accurate forecasting of wind speed is critical [10], [11]. In addition to its speed, wind is also characterized by its direction, and the time of occurrence. Wind energy is derived from natural wind flow depending on the force with which it moves or its speed. The successful profiteering of wind energy depends on the wind resources available in the area [12]. The economic viability of wind energy converters is determined by the wind conditions at a given location. Wind turbines require wind to be greater than 4 m/s so as to generate electricity [13]. Lower wind speeds can be sufficient for wind turbines. Most wind turbines, on the other hand, start to furl when the wind speed is between 12 and 15 m/s. It is difficult to