System Identification: - A Criterion, or Verification Function

Introduction
System identification is a methodology developed mainly in the area of automatic control.

It is meaning to choose the best model(s) from a given model set based on the observed input-
output data from the system.
The problem of system identification is specified by three elements
A data set D obtained by input-output measurement.
To obtain useful data for system identification a priori information or some physical knowledge
must be provided
A model set M , or a model structure, containing candidate models.
A choice of model set M is a difficult issue in system identification , but usually several class of
discrete-time linear time-invariant systems can be used.
The models which do not necessarily reflect the knowledge about the structure of the system are
referred to as block-box models.
By using some physical principles, a models contain several unknown parameters can be
constructed. These models are called gray-box models because some basic laws from physics
are employed to describe the dynamics of a system or a phenomenon
Find the model from models set M, by which the experimental data is best explained.
A criterion, or verification function.
to select the best model(s) or a rule to
evaluate candidate models, based on the measured data.
And to Verify how the model is close to real system .
This can be done by using the following criterion
N 1
V N l ( y (t ), y M (t ), u (t ))
t 0
Where l(.) is a nonnegative loss function, and N the number of data.
Given three basic elements in system identification, in principle it can find the best model
mM
to select the best model the following steps are necessary :
- define the condition for the existence of a model that minimizes the error .
- An algorithm for computing models must be .
- A method of model validation.
In particular, model validation is to determine whether or not an identified model should be
accepted as a suitable description that explains the dynamics of a system.
In this case, to select the best model follow the steps:
- A condition for the existence of a model that minimizes the criterion
must be provided.
- An algorithm of computing models.
- A method of model validation.
In particular, model validation is to determine whether or not an identified model should be
accepted as a suitable description that explains the dynamics of a system.
Model validation is used on the way in which the model is used. if the transfer function is
identified , the quality of an identified model is evaluated based on the step response and/ Or the
pole-zero configuration.
If the performance is not satisfactory, then the earlier stages of system identification, including
the selection of model structure , or experiment design must be redesign.
A flow diagram of system identification can be displayed in the following figure
flow diagram of system identification
where the system identification procedure has an inherent iterative or

feedback structure.
System identification technique definition
The System Identification problem is to estimate a model of a system based on observed
input-output data. Several ways to describe a system and to estimate such descriptions exist.
This section gives a brief account of the most important approaches. The procedure to
determine a model of a dynamical system from observed input-output data involves three basic
ingredients:
- The input-output data
- A set of candidate models (the model structure).
- A criterion to select a particular model in the set, based on the information in the data (the
identification method).
- The identification process amounts to repeatedly selecting a model structure, computing the
best model in the structure, and evaluating this model's properties to see if they are satisfactory.
- The cycle can be itemized as follows:
1. Design an experiment and collect input-output data from the process to be

identified.
2. Examine the data. Polish it so as to remove trends and outliers, and select
useful portions of the original data. Possibly apply filtering to enhance
important frequency ranges.
3. Select and define a model structure (a set of candidate system descriptions)

within which a model is to be found.
4. Compute the best model in the model structure according to the input-
output data and a given criterion of fit.
5. Examine the obtained model's properties.
6. If the model is good enough, then stop; otherwise go back to Step 3 to try
another model set Possibly also try other estimation methods (Step 4) or
work further on the input-output data (Steps 1 and 2).
The System Identification offers several functions for each of above steps.
For Step 2 there are routines to plot data, filter data, and remove trends in
data, as well as to resample and reconstruct missing data.
For Step 3 the System Identification offers a variety of nonparametric

models, as well as all the most common black-box input-output and state-
space structures, and also general tailor-made linear state-space models in
discrete and continuous time.
For Step 4 general prediction error (maximum likelihood) methods, as well

as instrumental variable methods and sub-space methods are offered for
parametric models, while basic correlation and spectral analysis methods
are used for nonparametric model structures.
To examine models in Step 5, many functions allow the computation and

presentation of frequency functions and poles and zeros, as well as
simulation and prediction using the model.
Linear Time-invariant System Models
Linear time-invariant (LTI) systems are the most important class of dynamical
systems . Although they represent idealizations of the processes encountered
in real life, the approximations involved are often justified, and design
considerations based on linear theory lead to good results in many cases .
The system is to be time invariant if its response to a certain input signal dose
not depends on absolute time, it is said to be linear if its output response to a
linear combination of inputs is the same linear combination of the output
responses of the individual inputs. Furthermore, it is said to be causal if the
output at certain time depends on the input up to that time only.
This chapter will provide a brief review of the basic models used in
representing LTI systems.
Rational Model Representation of Transfer-Functions :
The most immediate way of parameterize G and H is to represent them as rational

functions and let the parameters be the numerator and denominator coefficients. Such
model structures are also known as black-box models .
ARX-MODEL Structure:
( Auto Regressive with exogenous input)
The simplest input-output relationship is obtained by describing it as a

linear difference equation:
y(k ) a1 y(k - 1) ... ana y(k - n a )
(2.20)
b1u (k - 1) ... b n b u (k - n b ) e(k)
Where:
y (k - i) Output in previous instant.
u (k - i) Input in previous instant.
na Number of a coefficients.
nb Number of b coefficients.
e(k ) = is the white-noise
The white-noise term e(k ) enters as a direct error in the difference
equation.
Let:
[a1 a2 ... an a
b1 b2 ... bnb ]T (2.21)
A(q) 1 a1q -1 ... ana q -na

And
B(q) b1q -1 ... bnb q -nb
B(q) 1
Then G(q, ) , H ( q, ) (2.22)
A(q) A(q)
The model in Eq.(2.20) is called the ARX-MODEL, where AR refers to

the autoregressive part A(q ) y (k ) and X to the extra input B(q)u (k ) .
ARMAX-MODEL Structure:
(Auto Regressive Moving Average with exogenous input)
(2.25)
OE-MODEL Structure:
(Output Error model structure)
(2.28)
BJ-MODEL Structure:
(Box-Jenkins model structure)
(2.30)

System Identification: - A Criterion, or Verification Function

Uploaded by

Copyright:

Available Formats

System Identification: - A Criterion, or Verification Function

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

System Identification: - A Criterion, or Verification Function

Uploaded by

Copyright:

Available Formats

Introduction

System identification is a methodology developed mainly in the area of automatic control.

where the system identification procedure has an inherent iterative or

- The input-output data

- A set of candidate models (the model structure).

1. Design an experiment and collect input-output data from the process to be

3. Select and define a model structure (a set of candidate system descriptions)

5. Examine the obtained model's properties.

For Step 3 the System Identification offers a variety of nonparametric

For Step 4 general prediction error (maximum likelihood) methods, as well

To examine models in Step 5, many functions allow the computation and

The most immediate way of parameterize G and H is to represent them as rational

The simplest input-output relationship is obtained by describing it as a

A(q) 1 a1q -1 ... ana q -na

The model in Eq.(2.20) is called the ARX-MODEL, where AR refers to

You might also like