Resource-Efficient Widely-Linear Collaborative Wind Forecasting for Renewable Energy Generation

aalto university school of electrical engineering abstract of the master's thesis Wind is an important source of renewable energy with abundant availability at many places of the world. The integration of wind energy into the electricity grid brings new challenges on the system reliability due to its intermittent nature. A robust tool with good accuracy for wind power forecasting is an essential part of wind based power system. Wind power is directly related to the cube of wind speed and hence a small improvement in the forecasting of wind speed provides larger improvement in wind power forecasts. The accuracy of a wind speed forecasting model depends upon the characteristics of wind signal incorporated in the model. Wind direction has significant effect on wind speed forecasts and the complex representation of wind signal provides a convenient tool to include the effect of direction. The complex wind signal is non-circular, and widely-linear modeling of the complex wind signal provides optimal results. Wind signal has strong temporal and spatial correlations as well. Most of the wind forecasting models at present do not incorporate all of these characteristics of the wind signal. In this thesis work, we will develop a widely-linear collaborative wind forecasting model which takes into account the non-circular nature of the complex wind signal and its temporal and spatial correlations as well. The parameters of the model are updated using recursive least squares approach. The forecasting model is further optimized for efficient resource utilization by using a set-membership formulation. A data model for the complex wind signal is developed and the algorithm is tested on the simulated wind data. Simulation results show better forecasting performance with the developed approach. Real world wind data is also used to verify the results. Preface First of all, thanks to the Almighty Allah for the countless blessings bestowed upon me throughout my life. The thanks to my supervisor D.Sc. (Tech.) Stefan Werner are difficult for me to put in words. This thesis would have not been possible without his constant guidance, patience and encouragement. I acknowledge the constructive comments and reviews from my instructor M.Sc. (Tech.) Taneli Riihonen, which helped me to improve the structure and presentation of the thesis. Many thanks to my friends Aamir, Pramod, Ali and others for always being there to cheer me up. Last but not least, my family, whose constant encouragement and emotional support helped me to achieve all …