Due to the gradual deployment of wind farms at global scale, wind power comes with variability and uncertainty to the power grid network. Many challenges come with selecting the optimal wind farm for efficient wind power generation. Thus, these challenges come with intermittent behaviors such as wind speed and other weather parameters variation that lead to uncertainty in wind power generation. Due to the above reasons, an accurate prediction platform is essential for effective future wind turbine site selection. In this paper, we propose a software-defined quadcopter system (SDN controller) with neural network (NN) power prediction system stack. The SDN controller based on quadcopter manages a ground Mobile station that travels on a given set of way-points to measure the best optimal wind speed quality. We consider the ground mobile station as an open flow switch that collect sensor data and send it to SDN controller for aggregation and management. Our proposed SDN-NN stack is designed with multi-layered artificial neural network system that employees meteorological sensor data such as wind speed and wind direction to forecast the output generated power. The NN uses the feed-forward algorithm and implement the back propagation algorithm for tuning the weight values of the neural network. To demonstrate the effectiveness of our platform, a wind farm practical climatic data set was used for evaluation and training. The experimental results showed that our proposed system achieved 93% significant and reliable prediction accuracy in power generation when using multi-climatic variables rather than using one climate variable that provided 87% prediction only. A 6% improvement in prediction was made using our proposed platform.
|Publication status||Published - 21 Jan 2019|
|Event||2018 IEEE Global Conference on Internet of Things - Alexandria, Egypt|
Duration: 5 Dec 2018 → 7 Dec 2018
|Conference||2018 IEEE Global Conference on Internet of Things|
|Period||5/12/18 → 7/12/18|