The Application of Neural Networks for Spin Avoidance and Recovery

This paper presents a method by which artificial neural networks can be trained and used to identify a possible spin entry, differentiate between an incipient spin and a stabilized spin, and predict required recovery controls. These were then implemented into a simulation and tested using data from actual flight tests conducted by NASA Langley Research Center, to verify that artificial neural networks can successfully be used for this application. The spin avoidance and recovery system functioned properly. In addition, a weighting system was developed to predict possible spin characteristics of aircraft, depending on the relative magnitude of the three principal moments of inertia.