Neuro-fuzzy techniques for the identification of aircraft wake vortex encounters

摘要

The automatic identification of aircraft vortex encounters by soft-computing techniques, such as fuzzy systems, gives better identification results than when performed by subjective manual approaches. However, the construction of the fuzzy rules heavily depends on a thorough understanding of vortex aerodynamics and aircraft response to a vortex encounter. Moreover, it requires a detailed analysis of flight data records and pilot reports, which all help in the proper selection of the critical fuzzy input variables. However, the automatic identification of vortex encounters by fuzzy logic can be greatly influenced by the shape and distributions of membership functions. Artificial neural networks integrated with fuzzy systems have been used as a solution in the automatic tuning of the membership functions of fuzzy linguistic variables. In this paper, a hybrid Adaptive Neuro-Fuzzy Inference System (ANFIS) is selected to automatically tune the parameters of the fuzzy membership functions. Various neuro-fuzzy models having different sets of parameters and factors are investigated. Simulation results with an average identification accuracy of 84.2% is obtained. The neuro-fuzzy hybrid model can also be used in the identification of other aircraft flight events such as wind shear and hard landing.