Research On Sliding Mode Multivariate Control And Simulation Of Turboshaft Engine

Aero-engine is a kind of complicated nonlinear system, in the whole flight envelope, theoperating state of the engine changes greatly. Usually, a suitable mathematical model of the engine isrequired during the investigation on aero-engine control systems. Because of the complexity of theengine’s physical structure, and the unavoidable and uncertainties in the flight environment, etc, thereis a great difference between a simple mathematical model and a real engine. In this dissertation,aimed at turboshaft engine, a multivariable control scheme is carried out based on sliding modecontrol (SMC) theory which can make the closed-loop system has strong robustness. The detailedcontributions of this dissertation are given as follows:Firstly, by using the SIMULINK and C mixed-language program method, and according to thebasic principle of the engine, such as the quality of the law of conservation of energy, acomponent-level model of the engine is established, so that simulation can be done easily and quicklyon this model in the following work. Compared with the C language programming method or theMATLAB script programming method, the presented hybrid programming method can make thesimulation model obtains faster operation speed, and more facilitate for control systems design andanalysis.Secondly, an error correction based small deviation model is discussed. This method is simpleand the calculate speed is fast, however, the errors are slightly large. Therefore, another smalldeviation model based on self-optimizing genetic algorithm is applied, subsequently. Such a methodhas smaller error but is more computational. Hence, a nonlinear least-squares solution method is usedto calculate a small deviations model. Compared with the former two methods, this method provides atrade-off between computational complexity and model error. Then, based on the small deviationmodel, by using BP neural network method, a variable parameter model of turboshaft engine in a widerange of flight envelope is presented. The modeling method based on BP network is simpler inhigh-dimensional structure and is with smaller modeling error, by comparing with the modelingmethod based on linear interpolation. Therefore, the modeling method based on BP network is moresuitable for use in the simulation.Finally, according to SMC theory, by using optimization sliding hyperplane and based on twodifferent reaching law methods, two turboshaft engine controllers are obtained. The two reaching lawmethods are proportion-variable speed-constant speed (PVC) reaching law method and PID reaching law method, respectively. The simulation results show that, both SMC controllers can suppress theinfluence of the uncertainties from the outside or internal of the system. However, the gas turbinespeed and power turbine speed have better dynamic performance under the SMC controller on thebasis of PID reaching law method.