The computational efficiency of traditional model order reduction (MOR) methods may degrade sharply for multi-input multi-output (MIMO) systems especially when the number of ports of MIMO systems is very large. During the concrete computation process, many input-output pairs can be ignored due to the weak interactions to each other, and hence the efficiency of reduction can be improved by reducing the number of ports. In this brief, we develop a dynamic relative gain array (DRGA) method to decide which inputs are important enough to an output in the MOR process. The DRGA method is based on the state feedback predictive control, and both the steady state information and the dynamic information are considered in the process of loop pairing. Multi-input single-output (MISO) subsystems can be obtained from decoupling the original large MIMO system. Experimental results on RLC networks show that the proposed DRGA based MOR method has higher accuracy compared with the passive reduced-order interconnect macromodeling (PRIMA) method, the decentralized model order reduction (DeMOR) method, and the balance truncation reduction (BTR) method.

中文翻译：

---

基于动态相对增益阵列的MIMO系统模型降阶

对于多输入多输出（MIMO）系统，传统模型降阶（MOR）方法的计算效率可能会急剧下降，尤其是当 MIMO 系统的端口数量非常大时。在具体的计算过程中，很多输入输出对之间由于相互作用较弱而可以忽略不计，因此可以通过减少端口数来提高归约效率。在这个简介中，我们开发了一种动态相对增益阵列 (DRGA) 方法来决定哪些输入对于 MOR 过程中的输出足够重要。DRGA方法基于状态反馈预测控制，在回路配对过程中同时考虑稳态信息和动态信息。多输入单输出 (MISO) 子系统可以通过解耦原始的大型 MIMO 系统获得。