Research On Turboshaft Engine Modeling And Intelligent Control

Turboshaft engine is a time-varying system with high nonlinearity. Its real work environment isbad, and there are many uncertain factors and control variable over the full-envelope flight range. Inorder to do better research for the engine, it is very necessary to establish a precise mathematicalmodel of the engine, and then design the control system on the mathematical model. At present, theintelligent control composed by the artificial neural network control has strong self-learning andadaptive function, it can effectively solve the difficult problem with complex control object andnon-linear in the turboshaft engine.In this paper, to start with, a nonlinear component level model of the turboshaft engine isestablished adopting the analytic method based on the pneumatic and thermodynamics law. The realworking part of the engine is replaced by the corresponding pneumatic and thermodynamics equation.According to the characteristics data of the airscrew, the model of airscrew is established using thefitting method. All components are connected to form a whole using the conversation law of qualityand energy which is must obeyed on the engine work time, and at the same time the common workingequations is gained. Then, by using the Newton-Raphson method to solve these nonlinear equations,some related parameters of the engine are obtained, which can use for simulating the real workingstate of the engine when the pneumatic thermal state changes.At last, the control system is designed according to the established model of engine. Both thesingle neuron network and BP neural network have the ability of self-learning and self-adaption. Thatcombining the single neuron network and BP neural network with the traditional PID controller coulddesign the intelligent PID controller of single neuron network and BP neural network which canadjust parameters on-line. Both the controllers are designed based on power feedback. That takingpower directly into controller for participating in control enhances the system control sensitivitygreatly as well as improves the control efficiency. The simulation of dynamic and steady state isimplemented by using the both controllers so as to verify the correctness of the model and theperformance of the controller. Then, according to the simulation results, the intelligent optimizingcontroller is established by synthetizing the two controller features.