Modelling is important in studying systems. It shows the system from different angles, help in analysing and understanding the system, analysis the causes and effects, can discover the defects in the system and treat it, illustrates the alternatives and studies their effects. Research looks for appropriate models of low, medium, high and very high wind speed. ARMA can be used in making time series models. It can get models with similar properties of original data. This research develops several ARMA models for wind speed data classified according to its value in four groups. The least mean square error (MSE) is the criteria to choose between models, for each wind speed model gets the appropriate wind turbine model that tells the predicted output power. The original data is taken to its group according to its value and taken to its appropriate power model. Turbine stops for low and very high speed. MPPT is used for medium speed. High speed models need pitch angle control. The model is implemented to Zafranaa wind speed data. It is divided into three groups. There were no data points for fourth group, Max point was 17m/s. Third group has appropriate amount of data and most of data are in second group, Finally the implementation procedure has demonstrated the applicability of the preferred method