Research On Turboshaft Engine Component Method Modeling And Neural Network Control

Turboshaft aero-engine control technology is one of the key technologies in the field of aero-engine research and development,and it has always been a research hotspot in academia and industry.The research on turboshaft engine control technology needs to establish a mathematical model for the engine.Based on the model,various control methods are used to achieve effective control of the engine.At present,the nonlinear modeling and control methods for vortex engines have received extensive attention.In the field of nonlinear modeling,component methods and experimental methods are more commonly used.In the field of control methods,intelligent control represented by neural networks has gradually become a hot topic of nonlinear control.Therefore,this paper develops the turbulence engine component modeling and neural network control research.First,under the MATLAB/Simulink platform,a component-level non-linear mathematical model is constructed for a certain type of turboshaft engine using the component method.This method divides the engine into six independent sub-components.Each component mathematically abstracts the characteristics of the component according to the aerodynamic thermal laws that are met during engine operation.In order to make the components work in harmony within the flight envelope,this paper establishes the steady-state and dynamic-state working equations of the engine based on the component-level model,and uses the N-R method to iteratively solve the engine’s normal working conditions.The component-level model lays a solid foundation for the following research on the control method of turboshaft engine neural network.Secondly,the BP neural network is introduced into the turboshaft engine power turbine PID controller to form a BP neural network single-loop PID control method.Using BP network’s self-learning and consistent approach capability,the PID parameters can be self-tuned.Simulation experiments show that the control method satisfies the basic requirements for the normal operation of the engine in steady-state and dynamic-state,and verifies the effectiveness of the control method.This method has good power turbine constant-speed control accuracy,but the control curve of each variable has slight jitter.Some of the variables are slightly over-adjusted.Finally,a turboshaft engine cascade PID control method based on RBF neural network is proposed and designed.Based on the PID control of the power turbine,the PI control structure is added to the gas turbine to enhance the anti-disturbance capability of the control system.In order to avoid the disadvantages of the BP network easily falling into local extremes and slow convergence,it is easy to obtain a global optimal solution and learn.The fast RBF neural network is introduced into the PID and PI controllers respectively to further improve the auto-tuning effect of the controller parameters.Simulation experiments show that the control method also achieves effective control of the steady-state and dynamic-state of the engine;the control curves change more smoothly,which significantly eliminates small jitter and suppresses the occurrence of overshoot,but the power turbine constant-speed control accuracy has weakened.The BP neural network single-loop PID control method and the RBF neural network cascade PID control method all meet the engine steady-state and dynamic-state control requirements.They have their own advantages in terms of power turbine speed control accuracy and control curve stability respectively.Both of these two methods are effective and complementary turboshaft engine control methods.