IIR Filter Design methods

[Sandi2000]

IIR filters can be designed using different methods ,one of the most commonly used is via the reference analog prototype filter,but there is other method is model order reduction .
I want more information about these two methods, when we can use the analog prototype filter for design iir filter ,and when we use the model order reduction techniques.

Thank you in advance.

 

[FvM]

I'm not aware of "model order reduction" as a filter design method. Any reference?

 

[pancho_hideboo]

I want more information about these two methods.

There are two methods.

(a) H_A(s) -> H_D(z) ; Analog Prototyping
(b) Direct Synthesis of H_D(z)

Simply we can realize H_D(z) by (b) which can not be realized by (a).

Mathworks MATLAB provide three methods as (b).
(1) Direct Design
(2) Generalized Butterworth Design
(3) Parametric Modeling

See https://mathworks.com/help/signal/ug/iir-filter-design.html?lang=en#brbq5qb

 

[Sandi2000]

There are two methods.

(a) H_A(s) -> H_D(z) ; Analog Prototyping
(b) Direct Synthesis of H_D(z)

Simply we can realize H_D(z) by (b) which can not be realized by (a).

Mathworks MATLAB provide three methods as (b).
(1) Direct Design
(2) Generalized Butterworth Design
(3) Parametric Modeling

See https://mathworks.com/help/signal/ug/iir-filter-design.html?lang=en#brbq5qb

Thank you for the clarification ,but i have seen this site https://www.mathworks.com/help/control/ug/about-model-order-reduction.html , i have another question, there is a relationship between parametric modeling and model order reduction?

 

[pancho_hideboo]

i have another question, there is a relationship between parametric modeling and model order reduction?

I don't what you want to mean by "model order reduction".

I know "Reduced Order Model".
But it has no relation to IIR filter.

You might get "Reduced Order Model" in "Frequency Domain System Identification Toolbox" of MATLAB.

http://en.wikipedia.org/wiki/Model_order_reduction
http://www.math.ucdavis.edu/~freund/paper_bdf.pdf
http://kiwi.mit.edu/papers/Parametric-model-reduction-UQ-Bui-Thanh-Willcox.pdf

http://home.mit.bme.hu/~kollar/fdident/
**broken link removed**
http://mathworks.com/products/conne...-identification-toolbox.html?requestedDomain=

 

[Sandi2000]

I don't what you want to mean by "model order reduction".

I know "Reduced Order Model".
But it has no relation to IIR filter.

You might get "Reduced Order Model" in "Frequency Domain System Identification Toolbox" of MATLAB.

http://en.wikipedia.org/wiki/Model_order_reduction
http://www.math.ucdavis.edu/~freund/paper_bdf.pdf
http://kiwi.mit.edu/papers/Parametric-model-reduction-UQ-Bui-Thanh-Willcox.pdf

http://home.mit.bme.hu/~kollar/fdident/
**broken link removed**
http://mathworks.com/products/conne...-identification-toolbox.html?requestedDomain=

Thank you ,i will read it carefully .

 

[pancho_hideboo]

but i have seen this site https://www.mathworks.com/help/control/ug/about-model-order-reduction.html

This has no direct relation with IIR Filter Design method.

Why do you think this is a IIR Filter Design method ?

 

[Sandi2000]

This has no direct relation with IIR Filter Design method.

Why do you think this is a IIR Filter Design method ?

I agree with you,

There is many techniques in model order reduction like balance truncation BT, impulse response graimian, Hankel approximation ..and for find a reduced system there need to have ''original system'', in generally , the original system must be transformed to state space representation, with use the controllability , observability in a closed form.
That is what i have a problem and i want to understand them,how can apply the state space ,how can i use the observability ,the controlability..?

Why do i think this is a IIR Filter Design method?
because these techniques is based on approximation of an original filter which satisfies the given frequency response specifications and subsequently obtaining a lower order IIR.