Publication

Abstract : The need for clean and green power is increasing daily, to satisfy these Nuclear Plants, Solar Plants and Windfarms are on the up-rise. Windfarms become an obvious choice near coastal regions where the wind speed is adequate for the efficient and reliable working of Wind Turbines. Focusing on the needs of a smart city, it becomes necessary for the management of Windfarms to also become smart, with foreplanning, layout optimization and smart transmission. Since the output of windfarms is largely seasonal and depends on wind speed, the need to forecast wind becomes a necessity. The forecasting of wind using machine learning algorithms can help estimate future power generation. This work will compare the performance of various Machine Learning Algorithms in forecasting wind speeds and direction. Windfarms need to optimized to offset the wake effect caused by the overlapping wake regions of the windmills, an ill-placed windmill can cause loss of efficiency and capacity. To combat the loss in efficiency and power in Windfarms, this work will present ways to optimize the performance of Windfarms using methods like Genetic Algorithm, Particle Swarm Optimization, Nelder-Mead and Random Search. The optimization performed will be subject to constraints such as Turbine type, Layout, Proximity and seasonal wind trends. The work will also present a novel approach to optimization derived from the existing work presented and will compare their performance. Genetic Algorithm and Nelder-Mead-Genetic Algorithm yield the best results, while moderately computationally expensive, they can help increase efficiency by 1–1.3%, Nelder-Mead Particle Swarm Optimization can improve efficiency by 0.6% and Random Search by 0.36% in 50 iterations. To allow for uninterrupted power supply, Machine Learning models will be used to classify faults in High-power Transmission Lines to help predict the nature and condition of fault.