Development of Empirical Models From Process Data: - An Attractive Alternative

Development of Empirical Models
From Process Data

In some situations it is not feasible to develop a theoretical
(physically-based model) due to:
1. Lack of information
2. Model complexity
3. Engineering effort required.
An attractive alternative: Develop an empirical dynamic
model from input-output data Identification.
Advantage: less effort is required
Disadvantage: the model is only valid (at best) for the
range of data used in its development.
i.e., empirical models usually dont extrapolate very
well.
Black box models

step response easiest to use but may upset
the plant manager (size of input change?
move to new steady-state?)
other methods
impulse - dye injection, tracer

random - PRBS (pseudo random binary sequences)
sinusoidal - theoretical approach
frequency response - modest usage (incl. pulse testing)
on-line (under FB control)
2
Simple Linear Regression: Steady-State Model
As an illustrative example, consider a simple linear model
between an output variable y and input variable u,
y 1 2u
where 1 and 2 are the unknown model parameters to be
estimated and is a random error.
Predictions of y can be made from the regression model,

y u
1 2
where 1 and 2 denote the estimated values of 1 and 2,

and y denotes the predicted value of y.
Let Y denote the measured value of y. Each pair of (ui, Yi)
observations satisfies:
Yi 1 2ui i (7-1)
The Least Squares Approach

The least squares method is widely used to calculate the
values of 1 and 2 that minimize the sum of the squares of
the errors S for an arbitrary number of data points, N:
N N
S Yi 1 2ui
2
12 (6-2)
i 1 i 1
Fitting First and Second-Order Models
Using Step Tests
Simple transfer function models can be obtained graphically
from step response data.
A plot of the output response of a process to a step change in
input is sometimes referred to as a process reaction curve.
If the process of interest can be approximated by a first- or
second-order linear model, the model parameters can be
obtained by inspection of the process reaction curve.
The response of a first-order model, Y(s)/U(s)=K/(s+1), to

a step change of magnitude M is:
y t KM (1 e t / ) (5-18)
The initial slope is given by:
d y 1
(7-15)
dt KM t 0
The gain can be calculated from the steady-state changes

in u and y:
y y
K
u M
where y is the steady-state change in y.

Figure 7.3 Step response of a first-order system and
graphical constructions used to estimate the time constant, .
First-Order Plus Time Delay Model

Ke- s
G (s)
s 1
For this FOPTD model, we note the following charac-
teristics of its step response:
1. The response attains 63.2% of its final response
at time, t = .
2. The line drawn tangent to the response at
maximum slope (t = ) intersects the y/KM=1
line at (t = ).
3. The step response is essentially complete at t=5.

In other words, the settling time is ts=5.
Figure 7.5 Graphical analysis of the process reaction curve
to obtain parameters of a first-order plus time delay model.
There are two generally accepted graphical techniques for

determining model parameters , , and K.
Method 1: Slope-intercept method

First, a slope is drawn through the inflection point of the
process reaction curve in Fig. 7.5. Then and are
determined by inspection.
Alternatively, can be found from the time that the

normalized response is 63.2% complete or from
determination of the settling time, ts. Then set =ts/5.
Method 2. Sundaresan and Krishnaswamys Method
This method avoids use of the point of inflection
construction entirely to estimate the time delay.
Sundaresan and Krishnaswamys Method
They proposed that two times, t1 and t2, be estimated from a
step response curve, corresponding to the 35.3% and 85.3%
response times, respectively.
The time delay and time constant are then estimated from the
following equations:
1.3t1 0.29t2
(7-19)
0.67 t2 t1
These values of and approximately minimize the

difference between the measured response and the model,
based on a correlation for many data sets.
Estimating Second-order Model Parameters

Using Graphical Analysis
In general, a better approximation to an experimental step
response can be obtained by fitting a second-order model to
the data.
Figure 7.6 shows the range of shapes that can occur for the
step response model,
K
G s (5-39)
1s 1 2 s 1
Figure 7.6 includes two limiting cases: 2 / 1 0 , where the
system becomes first order, and 2 / 1 1 , the critically
damped case.
The larger of the two time constants, 1 , is called the
dominant time constant.
Figure 7.6 Step response for several overdamped second-
order systems.
Smiths Method
Assumed model:
Ke s
G s
2 s 2 2s 1
Procedure:
1. Determine t20 and t60 from the step response.
2. Find and t60/ from Fig. 7.7.
3. Find t60/ from Fig. 7.7 and then calculate (since
t60 is known)
16
Smiths Method
20% response: t20 = 1.85
60% response: t60 = 5.0
t20 / t60 = 0.37
from graph
t60 1 2 1
2.8 = 1.79

= 1.3 2 2 1
Solving, 1 3.81
2 0.84
1 2 Sum of squares
S 3.81 0.84 0.0757
NLR (=0) 2.99 1.92 0.000028
FOPTD ( = 0.7) 4.60 - 0.0760
17
18
Neural Network Modeling
19

Development of Empirical Models From Process Data: - An Attractive Alternative

Uploaded by

Copyright:

Available Formats

Development of Empirical Models From Process Data: - An Attractive Alternative

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

Development of Empirical Models From Process Data: - An Attractive Alternative

Uploaded by

Copyright:

Available Formats

Development of Empirical Models

From Process Data

Black box models

impulse - dye injection, tracer

Predictions of y can be made from the regression model,

where 1 and 2 denote the estimated values of 1 and 2,

The Least Squares Approach

The response of a first-order model, Y(s)/U(s)=K/(s+1), to

The initial slope is given by:

The gain can be calculated from the steady-state changes

where y is the steady-state change in y.

First-Order Plus Time Delay Model

3. The step response is essentially complete at t=5.

There are two generally accepted graphical techniques for

Method 1: Slope-intercept method

Alternatively, can be found from the time that the

These values of and approximately minimize the

Estimating Second-order Model Parameters

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.