Adaptive Inverse Control of a Generic Hypersonic Vehicle Based on Improved EMRAN

An adaptive inverse controller is proposed for the flight control of a generic hypersonic vehicle. Other than conventional adaptive inverse control which adopt NARX algorithm, the proposed controller only use the input data and the change of the states to identify the inverse of plant. The improvement can reduce the input and output of the adaptive inverse filter, and improve the real-time performance. To accelerate the convergence of the adaptive inverse filter, original EMRAN algorithm is improved. By allocate and combine the error variance matrices, the number of hidden neurons be updated in each training cycle can be determined according to the calculation capacity of the platform, and both the convergence speed and accuracy can be increased without losing of real-time performance. Simulation result under a typical flight condition is presented to demonstrate the improvement.