一种求取发动机状态变量模型的改进拟合法

采用的拟合辨识方法是直接对根据状态变量模型中各变量间的线性关系对系数矩阵进行拟合，并且实现对稳态过程和瞬态过程进行分步拟合，从而消除瞬态误差对稳态精度的影响．本方法有接近传统最小二乘法的精度，并且无需求解状态变量模型微分方程或使用迭代算法，只需要进行简单的矩阵运算即可得到各系数矩阵，因此计算效率远高于传统最小二乘法．同时，该方法也可与其他拟合方法结合使用从而进一步提高拟合精度．最后，应用该方法建立了某涡轴发动机的小偏差状态变量模型，通过与非线性模型仿真结果比较，验证了该方法的有效性，即使对象非线性较强，该方法也能得到精度较高的拟合效果．

An improved method of identification for aero-engine's state variable model

The method for aeroengines' state variable model(SVM)identification was implemented by using the linear relation of the variable of the differential equations,and separating the steady process and transient process to eliminate the steady errors coupled by transient process. Without solving the differential equations or using any iterative methods, the coefficient of the SVM could be acquired simply using matrix operation, so it has a higher computational efficiency than former least square method. And this method might also be combined with other identification method to realize better effect. Finally, to illustrate the validity and accuracy of the proposed method, a small perturbation SVM of some turbo-shaft engine model had been established through it, and the simulation results from the engine's state space model and nonlinear model were compared,and we can draw a conclusion that the SVM got by this method meets the dynamic and static objectives well, though the plant has a severe nonlinearity.