Research On Simulation And Control System For Micro Turbine Engine

The modeling and real-time simulation controlling system for a micro turbine engine (MTE)could possesses many advantages, such as reducing engine testing expenses, shortening developmentperiod, decreasing the cost, avoiding the rig-test risk and so on. Recently, Matlab is a popularnumerical simulation software and Simulink is one of the most important components in Matlab, itprovides an integrated environment of dynamic system modeling, simulation and comprehensiveanalysis. The modeling and control algorithm of MTE by Matlab are studied in this paper.Firstly, based on the engine test data, a nonlinear component-level model for the studied MTEwas developed with component characteristics, and then the model based on C-language wasprogrammed to S-function. The S-function model for MTE was developed on Simulink at last.Secondly, The Radial Basis Function (RBF) neural network and LS-SVM identificationmodeling method is studied and validated by the measured data of the MTE. The properties of themodels were researched and compared, which shows the LS-SVM model is higher accuracy and canbe used in the controlling system research.Finally, for the speed control of MTE, a single-step predictive control algorithm based on leastsquare support vector machine (LS-SVM) model and velocity variation particle swarm optimization(V-PSO) was presented. The problems such as the found of the LS-SVM predictive model, the rollingoptimization of the control values and combination of the predictive model and the controller in thecontrol loop were studied. The simulation of the MTE speed control system indicates that the steadystate control of MTE can achieve good performance with this manner. When deviated from designpoint, the parameter of the LS-SVM predictive model will be modified on-line with respect to its ownstrong learning and adaptive abilities, which guarantee the closed-loop control system maintains goodperformance and strong robustness.