Transformation model of thrust-vectoring using RBF neural network

Abstract

In this paper, a transformation model between the thrust-vectoring vane deflections and the resultant thrust deviation angles is established for the thrust-vectoring with three vane construction based on the radial basis function (RBF) neural network. The RBF neural network is trained using the experiment transformation data from NASA research memorandum via the generalized growing and pruning algorithm (GGAP). The established RBF neural network model can eliminate the inaccuracy of existing estimation model and avoids the modeling difficulties using the experiment data. To test the correctness of the transformation model using RBF neural network, it is compared with the existing estimation model. Through the simulation results, one can obtain that the RBF neural network transformation model established in this paper has a global and accurate description for the transformation relationship between the thrust-vectoring vane deflections and the resultant thrust deviation angles. Moreover, it can show the characteristics of the thrust-vectoring more precisely.