Inverse adaptive neuro-control of a turbo-fan engine

Neural Networks have been successfully used for implementing control architectures for different applications. In this work we examine neural network augmented inverse adaptive control of a turbo-fan engine. This implementation is classified as a level one intelligent control where changes in the system during the course of its operation can be corrected with an adaptive controller. The inverse control architecture presented uses a neural network to augment an existing proportional-integral (PI) controller to accommodate engine changes. The present architecture is implemented on a linear model as well as on a nonlinear engine model that was provided by General Electric. The non-linear engine model used for the simulation is a XTE46 turbo-fan engine. The model simulates a changed engine by changing the flow and efficiency scalars of the various components of the engine, namely, the fan, the compressor, and the turbines. Results of using the inverse adaptive neuro-control show excellent command following abilities even in the presence of big engine changes.