Abstract
In this paper, a new state space design method of filtered inverse system for linear time-invariant systems is proposed. Ordinary inverse systems are classified two types.
1) Inverse systems which make a indentity transfer function matrix of tandem union with original system.
2) Inverse systems which make a transfer function matrix integral value of identified matrix. But both types require differentiators to estimate unknown inputs.
This problem was solved by the idea of filtered inverse systems, in which the input is reproduced appoximately. But the filtered inverse was designed via transfer function matrix, and the computation was complex.
We present a new state space design method of filtered inverse system. This method is based on the fusion of observer and decoupling method. There are many advantages such as a simple structure and easy computation.
