Question about ARMA process: Changing a Stochastic Process into a Transfer Function

Click For Summary
SUMMARY

The discussion focuses on converting an ARMA process into a transfer function. A participant suggests that the initial equation presented is incorrect due to misalignment in the powers of the variable z, specifically indicating that the right side contains single-step signal delays of 1, 2, and 3. The correct formulation is provided as W(z)(1-1.08z-1+0.81z-2)=V(z)(z-1+0.1z-2-0.9z-3), leading to the transfer function W(z)/V(z) = (z2+0.1z-0.9)/(z3-1.08z2+0.81z).

PREREQUISITES
  • Understanding of ARMA processes in time series analysis
  • Familiarity with transfer functions in control theory
  • Knowledge of z-transform and its applications
  • Basic mathematical manipulation of polynomials
NEXT STEPS
  • Study the properties of ARMA processes and their applications in forecasting
  • Learn about the z-transform and its role in signal processing
  • Explore the derivation of transfer functions from stochastic processes
  • Investigate the implications of single-step delays in system modeling
USEFUL FOR

Researchers, engineers, and students in fields such as control systems, signal processing, and time series analysis who are looking to deepen their understanding of ARMA processes and their conversion to transfer functions.

e0ne199
Messages
56
Reaction score
5
TL;DR
I want to change arma process into transfer function, I have done it but I am unsure if it is correct or not
hello everyone, I have a question about stochastic process (ARMA process) that looks like this :
1728286558931.png

I would like to change it into a transfer function, so the final result looks like this :
1728286499447.png

My question is, is this equation correct? if it is not correct, what should I change for this equation? any response is really appreciated, thx.
 
Last edited:
Physics news on Phys.org
I think that's not exactly right, but it's a good try. The right side has single-step signal delays of 1, 2, and 3. So your powers of ##z## (on one side or the other) are off by 1. (I am assuming that the index in your variables are related to time such that k-1 is a step backward in time, a one step delay, ##1/z##)
I am not an expert at this, but here is what I would do. I prefer to work with the ##z^{-1}## single-step delay, rather than the ##z## look into the future, but that is just a matter of preference. I would get:
##W(z)(1-1.08z^{-1}+0.81z^{-2})=V(z)(z^{-1}+0.1z^{-2}-0.9z^{-3})##
So ##W(z)/V(z) = (z^2+0.1z-0.9)/(z^3-1.08z^2+0.81z)##.
 
Last edited:
  • Like
  • Love
Likes   Reactions: PhDeezNutz and e0ne199
FactChecker said:
I think that's not exactly right, but it's a good try. The right side has single-step signal delays of 1, 2, and 3. So your powers of ##z## (on one side or the other) are off by 1. (I am assuming that the index in your variables are related to time such that k-1 is a step backward in time, a one step delay, ##1/z##)
I am not an expert at this, but here is what I would do. I prefer to work with the ##z^{-1}## single-step delay, rather than the ##z## look into the future, but that is just a matter of preference. I would get:
##W(z)(1-1.08z^{-1}+0.81z^{-2})=V(z)(z^{-1}+0.1z^{-2}-0.9z^{-3})##
So ##W(z)/V(z) = (z^2+0.1z-0.9)/(z^3-1.08z^2+0.81z)##.
hahaha thank you, that explanation is what I have been missing until now 😁
 
  • Like
Likes   Reactions: PhDeezNutz, berkeman and FactChecker

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 9 ·
Replies
9
Views
1K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 13 ·
Replies
13
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 24 ·
Replies
24
Views
4K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 5 ·
Replies
5
Views
10K