Effects of missing data imputation by lineer interpolation and ARIMA models on energy calculation in wind speed series

Wind farms as a nature of the bussines are located in areas where there are tough air conditions to gather high power outputs with use of high wind speeds. It can be seen as an advantage but in measurement phase there might be data losts because of the tough air conditions. It is possible to use different methods in order to eliminate data losses in measurement logs. In this thesis recovery of lost data in wind speed measurements with use of mathematical prediction methods and calculation of power output with the new data which includes predicted values examined. Actual wind speed measurements of Zonguldak Devrek Meteorology station is used for the study. To create a data log which includes lost data in it, 20% of the real data, which is the maximum amount allowed to be lost deleted in the real measurements. To have a data log with lost datas in it, first randomly distrubuted datas deleted from the original measurement. For the second study bigger data blocks deleted to simulate periodic lost datas. For two different logs with lost datas, there had been two different calculations. First lost datas calculated as they change linearly. Second lost datas calculated with help of ARIMA method. Power outputs of a selected wind turbine for the area recalculated for each cases. In the end all four power output calculations compared to each other and power output for the original measurement to see if they similiar with the original result.