Research On Aeroengine Control On Wavelet Neural Networks And Its Experimental Verification

Multivariable adaptive PID control scheme is studied for a turbofan engine in this thesis. PID control method based on wavelet neural network and neural network PID control method based on wavelet neural network are proposed for a turbofan engine. Software has been designed for the Hardware In-Loop simulation platform. A research is done by this software in the Hardware In-Loop simulation platform.Benefited from wavelet transform being constrictive and fluctuant, it shows excellent temporal-frequency localization property, while it possesses such merits as ability of mapping nonlinear systems, self-learning, self-adaptation and so on. Thus, this network converges quickly with high precision and good robustness. This thesis adopts wavelet neural network as identifier , adjusts the parameters of the PID controller on-line, and gets the good results. At last, we use some examples to vertify the valid of the methods above.Using a certain turbofan aeroengine model as study object, the controllers are designed within the fly envelop with the control method above. To build the Hardware In-loop simulation platform, we use a computer as model, a computer as controller, C8051F021 single-chip and its relative circuit as sensor. The definition of data interface and the communication between signals are accomplished. We use Visual C++ 6.0 to develop the software of model and controller, and develop the software of C8051F021 single-chip by Silicon Laboratories IDE. Some Hardware In-loop simulations have been conducted to test the research results. The simulations do show that these results really have practical values.