基于优化拟形的航空发动机控制器降阶方法

Model reduction method based on optimization shaping for aero-engine controller

提出了一种基于优化输出拟形的航空发动机多变量控制器降阶方法,并以LQG/LTR控制器降阶为例,成功地将原控制器从8阶降为3阶。首先将控制器分为积分环节部分和稳定子系统部分,采用输出拟形的方法用低阶系统去逼近稳定子系统使二者具有相近的阶跃响应,利用遗传算法优化出低阶系统的参数,将低阶系统和分离出来的积分环节合并后即得到降阶后的控制器。与采用平衡截取Schur降阶方法比较,本方法可将系统阶次降得更低并取得更好的动态响应效果。

The model reduction method for aero-engine controller based on the optimization output shaping is presented in this paper. Using the LQG/LTR controller as an exampie, the 8^th order LQG/LTR controller is successfully reduced to a 3^th order system with this method. First the controller is divided into the integral part and the stable part. A lower order system is used to approach the stable part. With the optimization output shaping meth od, the lower order system parameters can be calculated using genetic algorithm （GA）. Then the step response of them is very close to each other. At last, combining the optimized lower order system with the separated integral part, the reduced order controller can be obtained. Compared with the balanced truncation Schur model reduction method, this method can.make the system with lower order and get better dynamic performance.