The Power of Data Towards the Optimization of the Wind Energy Exploitation by Panagiotis Korkos
Updated: Dec 2, 2019
The wind energy market is growing rapidly and environmental consciousness even more so. Climate change threatens the world as we know it and humankind is at a turning point. This creates the need to take important decisions to ensure a more sustainable future.
Scientists and engineers are at the forefront of activities which attempt to change this situation. One of the major areas of interest which is directly related to climate change is power generation.
The use of coal has deteriorated the situation contributing to global warming, thus the option of renewable energy sources seems imperative. The exploitation of solar and wind energy can make the change that is needed. Especially concerning wind energy, wind turbine technology is so advanced that it will be able to power entire countries in the future.
Finland has gradually increased the installation of wind turbines. According to latest information of 2018, nowadays 6.7% of Finland’s electricity consumption is covered by wind energy.
In 2019 the power generation from wind energy will be increased by 20 %. This will be possible by installing new wind turbines, especially in the region of Ostrobothnia. The objective which has been set by Finland is that by 2030 wind energy will provide 30% of total electricity consumption.
Wind energy and generally renewable energy sources can make the use of electric vehicles and boats more sustainable, as the power would not be generated at a coal factory. Furthermore, data centers, which are nowadays vital, require a lot of power to operate and this power can be provided using renewable energy.
A very important case towards this direction was the agreement that Google signed with a Finnish company to power up its data centers in Finland through wind farms. Moreover, there is ongoing research regarding smart grids, in which interconnected sources of energy like wind turbines, solar panels, power generation facilities and even electric vehicles and boats can balance the grid through the appropriate processing of data.
Nevertheless, wind turbine technology faces a big challenge, which is the big data coming from a quite large amount of sensors whose task is to record wind turbine’s operation 24/7 . Within these data, there is useful information which can improve its operation, reduce costs and make predictions for failures in order not to pollute the surrounding environment in the case of a catastrophe.
This challenge is addressed to engineers and it can be solved by using advanced signal processing techniques, especially machine learning and generally Artificial Intelligence (AI). AI provides great potential in order to not only detect signal anomalies but also identify these failures.
This way technicians would get correct information so that they know for instance what component to fix or replace, thus reducing the cost of unnecessary replacements. Additionally, it is claimed that in the near future data will be as important as the labor and the capital in a company.
The need for using data from wind turbines has been highlighted in the case of offshore wind turbines. The number of turbines in these areas is increasing rapidly and there are a lot of costs generated concerning the maintenance (e.g. boats to go on site, extremely qualified personnel for maintenance, bad weather conditions, safety issues, risk of natural disasters, etc.).
Furthermore, research has been conducted recently regarding power generation predictions using artificial intelligence techniques and statistical processes. Among them, Google and more specifically the Deepmind team developed a system which can predict the power output one day beforehand, thus balancing the grid and regulating demand.
Moreover, statistical processing provides the opportunity to model the power prediction and this issue attracts a lot of researchers. ARMA (Autoregressive - Moving Average) models are used to create models of each measured parameter and this option seems promising.
These models are capable of efficiently predicting the behavior of each component and they are very powerful, though computationally expensive. Another interesting application of AI in wind turbines is the use of a wavelet based fully convolutional neural network (FCNN) to detect ice in turbine blades. This problem is very severe and can stop the operation of wind turbines.
1. About wind power in Finland, https://www.tuulivoimayhdistys.fi/en/wind-power-in-finland/wind-power-in-finland, last retrieved: 18/8/2019
A. Kauranen, C. Steitz, S. Jacobsen, A. Smith, 2018, Google buys into new Finnish wind energy in renewables search, https://www.reuters.com/article/us-alphabet-renewables-finland/google-buys-into-new-finnish-wind-energy-in-renewables-search-idUSKCN1LR1OG, last retrieved: 18/8/2019
2. S. Kolumban, S. Kapodistria, and N. Nooraee, “Short and long-term wind turbine power output prediction,” ArXiv e-prints, 2017
3. S. Witherspoon, W. Fadrhonc, 2019, Machine learning can boost the value of wind energy, https://blog.google/technology/ai/machine-learning-can-boost-value-wind-energy/, last retrieved: 18/8/2019
4. B. Yuan, C. Wang, F. Jiang, M. Long, P. S. Yu, Y. Liu, “WaveletFCNN: A Deep Time Series Classification Model for Wind Turbine Blade Icing Detection”, ArXiv e-prints, 2019