抄録
Recently with the enlargement of plant's scale, model reduction is becoming much more important and many methods are studied in control system design.
The methods of order reduction for linear systems are divided into two groups. One treats transfer functions and the given high-order transfer function is approximately reduced to a low order one. On the other hand, the second method reduces its state space model by state aggregation. The latter method would be preferable in many practical viewpoint.
At present, many reduction techniques have mainly dealt with continuous-time systems and there are few for discrete-time systems.
Considering these situations, this thesis aims at proposing two reduction methods for linear discrete-time systems via state aggregation. We established the following methods:
i) Reduction based on discrete-Schwartz form,
ii) Reduction based on discrete-Routh form.
i) is a time domain approximation method, while ii) is a low-frequency domain one. Both of them assure the stability of reduced-order systems, and are performed by simple algorithm.
