Short-term wind power forecasting using the LMS algorithm

Abstract

-This paper presents the application of the LMS adaptive filter for the short-term wind power prediction, generated by a wind energy conversion system based on historical wind speed and wind power data, for a period of 24 hours with a step of one second. To compare the predicted power with that measured for a specific period. There are two steps in the process of predicting of wind power. In the first step, the raw data collected by the laboratory computer system. This step prepares the validated data to be used to construct an estimated model. In the second step, the LMS adaptive filter is applied to construct a model to predict wind power.Keywords: LMS algorithm, prediction, adaptive, wind power, turbine, renewable energy. I. INTRODUCTION The world's primary reserves of polluting primary energy (oil, natural gas, coal, uranium) sooner or later it will be exhausted, while the very industrialization of the last decades its demand for electricity is always increasing. This has pushed countries towards the use of nuclear power plants to meet its energy needs. This energy source has the distinct advantage of not causing air pollution unlike thermal power plants, but the risk of nuclear accidents, waste treatment and burial are real problems that make this energy unattractive to the future generations.Faced with this problem several countries have turned to renewable energies. Indeed, a real global challenge is taken seriously today, as well on the policy of reducing greenhouse gas emissions. These sources of renewable energy include wind energy and the one with the highest energy potential.It is clear that Morocco is seriously engaged in this challenge with the installation of several photovoltaic power stations and wind farms whose objective is to achieve a 50% renewable energy production of national production by 2030 , Which also motivated us to direct our research work on renewable energies, specifically wind energy.The main problem of wind power is that the primary energy (the wind) is not controllable. This means that the power output of the wind turbines will vary according to the wind speed. It is therefore necessary to implement solutions in order to allow the manager to limit the impact of these productions on electric network [1].Therefore, the objective of this work is to implement an efficient prediction method based on adaptive filters to provide an estimate of the output power of a wind turbine for prediction horizons not exceeding the hour. This will enable the manager in the future to optimize the management of his electricity network by facilitating the integration of the wind resource [2].II. PROBLEM One of the main factors contributing to slowing down the integration of wind energy into energy distribution networks is the variation of the power produced. Indeed, the energy production of a wind power plant depends on weather conditions that vary randomly. It is therefore impossible to control the power produced by the wind turbines [3]. Operators of the grid power which must meet the ene...