一种近似广义奇异摄动模型降阶算法

A Quasi-generalized Singular Perturbation Model Reduction Method

分析了平衡截断法、平衡奇异摄动法和广义平衡奇异摄动法存在的优缺点,提出了一种近似广义奇异摄动模型降阶算法。新算法在广义平衡奇异摄动法降阶过程中引入一个实参数代替矩阵,不仅集成了平衡截断法和奇异摄动法的优点,而且减少了计算量,大大缓解了利用广义奇异摄动法得到的低阶系统可能面临的无法实现的问题。仿真结果表明方法简单有效。

The advantages and disadvantages of balanced truncation, singular perturbational method and generalized singular perturbation are analyzed. A quasi-generalized singular perturbational model reduction method is proposed. This method not only combines the best features of balanced truncation and singular perturbational method, but also reduces the possibility of physical un-realization of the reduced model obtained from generalized singular perturbation method, according to a parameter chosen in the model-order reduction process. Typical simulations show the effectiveness of the proposed method.