The problem our solution solves
Wind power curves mainly contribute to wind power forecasting wind turbine condition monitoring, estimation of potential wind energy and wind turbine selection. Accurate prediction of wind power is critical to increasing the utilization of wind in the electricity grid. It also helps power system operators to plan unit commitment, economic scheduling, and dispatch. In general, an accurate power curve is conducive to wind power prediction. So, a suitable power curve results in more accurate power forecasts. Accurate estimation of wind energy potential is not only an essential part of wind energy development and utilization, but also provides investors with the necessary confidence in financial feasibility and risk mitigation. The wind power curve characterizes wind turbine generation under normal conditions, so it can be used as an online wind turbine power generation profile. This can help us understand whether a wind turbine works under normal conditions, and then allowing for troubleshooting and scheduling maintenance, as well as repair interventions when the turbine is faulty. Understanding potential failures will help maintain and improve the operational efficiency and reliability of wind energy conversion systems.
Challenges we ran into
In the initial process we faced problems such as selecting the proper dataset and the appropriate dataset that would help us prove our point. After that we found out that our dataset had many outliers present in it and we have taken care of that by defining 68 dataframes each having a distinct range of values of ActivePower output generation. After this we had to invest a chunk of time in properly training our model to maximum accuracy. We have got the accuracy of 0.99 in all the three prediction models. Since the accuracy was similar in all the three models we had a hard time choosing one as a final and more reliable model. Hence, for that we have also calculated the rms error value and selected the one with less value.