International Journal of Aviation Sciences (IJAS)


Avionics can be thought of as the nervous system sustaining aircraft operations, and accurate forecasting is critical in predicting system demand. Modeled after their biological counterpart, artificial neural networks (ANNs) maintain the ability to recognize trends in chaotic and nonlinear systems. These models possess the capability to produce more accurate predictions than those of traditional forecasting methods. The purpose of this paper is to provide a comprehensive overview of ANNs and detail the use of these techniques as a tool for forecasting in the avionics field. Furthermore, it introduces and examines several commonly used types of predictive ANNs. The performance of ANNs is compared to the Delphi technique and traditional time series forecasting techniques (simple moving average, exponential smoothing, and regression) to address aspects of both qualitative and quantitative forecasting. The paper also discusses recent and potential future advances in ANNs and the effects these applications may have in the field.

Included in

Aviation Commons