中文摘要：

针对含大规模输入输出端口的MEMS结构，提出分块叠加Amoldi方法提取其宏模型，针对MEMS器件中广泛存在几何形状相同而初始方位不同的结构，采用矩阵坐标变换实现宏模型的角度参数化，解决宏模型的重复提取问题。最后，结合一种微型可编程光栅的设计验证了宏模型的效率与精度，与有限元仿真结果进行比较，基于宏模型的MEMS器件系统级仿真结果相对误差小于2%时域仿真速度提高了45倍。

英文摘要：

A novel method of the angular parameterized macromodel extraction for the microstructure with a large number of terminals was proposed. First, the macromodel was extracted by combining the block Amoldi algorithm with the superposition theory for linear systems, and then the equivalent coordinate transformation was applied for angular parameterization of the macromodel. The angular parametrization could avoid the macromodel extraction repeatedly for those microstructures with the same geometry but different initial orientation. A micro programmable grating was used to demonstrate the proposed method. Numerical simulation results show that the macromodel can dramatically reduce the computation cost while capturing the device behavior faithfully. Compared with the FEM results, the relative error is less than 1.3%, and the computational efficiency for transient analysis improves about 45 times.