Dynamic characteristics identification and nonlinear regulator synthesis of a small gas turbine engine based on neural networks

Latterly, neural networks have been used to simulate gas turbine engine dynamics and regulator synthesis. However, little attention has been paid to structure rationalizing  of the neural network for identification problems. In addition, neural networks are often used for control issues. Usually these problems are tuning PID regulator coefficients or control in the certain modes. Therefore, it is important to study of the neural network structure dependence on the simulated engine parameter and the neurocontroller synthesis to control the engine in all modes. The neural network architecture was studied for rotor frequency parameter based on the engine tests. In the result neurocontroller was synthesed based on the JetCat P-60 SE engine model taking into account the limitations of the engine fuel consumption. The results allow us to reduce the total time to model engine and synthesys nonlinear controller.