用于离散时间模型降阶改进的最小信息损失方法

Revised Minimum Information Loss Method for Discrete-Time Model Reduction

针对最小信息损失方法进行模型降阶时结果不惟一的问题,提出了用于离散时间系统模型降阶改进的最小信息损失方法。本方法通过限制系统采用输出正规模型,将系统的能观性格兰姆矩阵限制为单位矩阵,从而使系统的总信息损失达到最小,保证了降阶结果的惟一性,并通过仿真进行了验证。

The RMIL（Revised Minimum Information Loss） method for diecrete-time model reduction is proposed to revise the MIL（Minimum Information Loss） method, whose result is not unique. By restricting the system to be the output-normal model and transforming the observability grammian to be an identity matrix, the present RMIL method causes the total information loss to be minimized and preserves the reduced-order model to be unique. Examples are also given to illustrate the approximating performance of the reduced-order model derived by RMIL