Research On Integrated Optimization Control Of Intelligent Aero-Engine

With the development of the modern aero-engine technology and the complexity of its operational environment,the requirements of the engine control technology are increase rigorous.Therefore,the IEC(intelligent engine control) is proposed and researched.The main characteristic of IEC is that the engine can adaptively operate in various control modes,such as baseline mode,life extending mode,emergency mode and performance mode,based on different environments and mission requirements.The optimization control is the key technology for the multi-mode control of intelligent engine.The theories and technologies of integrated optimization control for intelligent engine are researched here.First,the real time simulation models related to the optimal control for intelligent engine are developed.The aircraft / engine integrated real-time simulation platform is developed for simulating the dynamic and steady state of the engine under different flight tasks and flight conditions.For the steady model of turbofan engine,a novel modeling method based on simplex B-splines is proposed.This method has strong fitting ability and good generalization ability for smooth objects.For the dynamic model of turbofan engines,a new modeling method based on MGD NN(Mini-batch Gradient Descent Neural Network) is presented here.The MGD NN can be applied to engine model modeling with large flight envelope,multi-state and multi-control variable,which has large training data.The accuracy and generalization performance of the model are greatly improved by MGD NN.The accuracy and effectiveness of these modeling methods are verified by the simulation results.Secondly,the optimization control methods for engine acceleration are studied.To improve the response speed of the engine,a new optimization trontrol based the guide vane angle adjustment and improvement feasible sequential quadratic programming is proposed.The influence of guide blade angle adjustment of fan and compressor on the acceleration response speed of the engine is focused and studied.In view of the traditional acceleration optimization control only considers the optimal engine performance in a local period,a noval global optimization control,which comprehensive consideration the effects of control input in the initial and later period on engine response speed,is presented to improve engine thrust response performance.The time consumption for control variables stabilization is selected as optimization function.And the curves of control variables are selected as optimization variable.Thirdly,the researches of baseline model are carried out.The nonlinear model predictive control method based on the composite prediction model is proposed and applied to the baseline mode.The baseline mode mode chooses thrust as the control objective,which overcomes the shortcomings of traditional controller that controls thrust indirectly throungh engine measurable parameters,where the thrust is inaccuracy controlled.For improving the real-time performance of predictive control,a full envelope on board composite prediction model is established which improves the estimation accuracy of engine health parameters.The effectiveness and real-time performance of the control method is verified by the simulation results.Then,the researches on the extending life mode and the emergency mode are carried out.In extending life mode,a real-time optimization control method is proposed to extend turbo-fan engine service life.Thermal mechanical fatigue life model of turbine blade and the optimization model of life extending control are established.The main researches contents is how to reduce the thermal mechanical fatigue in the engine acceleration process.In emergency mode,a real-time optimization control method is proposed to enhance engine thrust response and enlarge its maximum thrust.Two emergency engine control modes,the overthrust mode and the faster responsemode,are redesigned.The influence of adjusting compressor guided vane angle control and broadening operation restrictions on the response performance and the output thrust are studied.Finally,the researches on performance model are carried out.The maximum thrust mode,minimum specific fuel consumption and minimum turbine inlet temperature are established.To improve the accuracy of on-board model,a new modeling method based on deep neural network is presented.Compared with the traditonal neural network,this modeling method has more strong fitting capacity because of the using of the deeper network structure.For improving the global optimization searching ability of the algorithm,a new hybid optimization method-genetic algorithm particle swarm optimization feasible sequential quadratic programming,is presented and applied to performance seeking control.The proposed method has the global search ability of intelligent optimization and the local search ability of mathematical programming algorithm.