A numerical model based on prior distribution fuzzy inference and neural networks

Growth in electricity demand gives a rise to the necessity of cleaner and safer electric supply and short-term wind speed prediction with high precision is irreplaceable in the efficient management of electric systems. However, it is both a challenging and significant task to achieve the accurate prediction of short-term wind speed. Many models lack stability and ignores the importance of meteorological factors, which leads to poor prediction accuracy. This paper develops a reliable numerical model for verification based on fuzzy inference (prior fuzzy inference network and adaptive network-based fuzzy inference system) and meteorological factors (atmospheric temperature, atmospheric pressure and atmospheric density). The fuzzy neural networks are a favorable scheme in wind speed predictions mainly due to their endogenous capacity of robust modeling of data sets with highly non-linear relationship between inputs and outputs. Three experiments covering the data collected from Hebei are performed to verify the effectiveness of the proposed hybrid model by comparing it with three well-known methods. It is concluded that the hybrid models proposed not only can satisfactorily approximate the actual value but they also can be an effective tool in the planning and dispatching of smart grids.