Scribd
Upload
7 views

Assign 2 SSP

Uploaded by

Shimul Das
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
7 views

Assign 2 SSP

Uploaded by

Shimul Das
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Statistical Signal Processing Assignment # 2

1. The simplified model of a DC motor is given in the following figure where va (t), ia (t), Ra and
La are the armature voltage, current, resistance and inductance, respectively. Jm is the motor
moment of inertia, and Kb and Kt are the back emf and motor torque constants, and Tm (t),
vb (t) and ω(t) are the motor torque, back emf voltage and motor speed, respectively.
Suppose thatThe simplified
the model ofRaaDCand
parameters motor
Kist given in the following
are known: Ra =figure Ω and Kt = 6 ∗ 10−5 N m/A.
0.2 where

𝑣 𝑡 + 𝟏 𝑖 𝑡 𝑇 𝑡 𝟏 𝜔 𝑡
𝑲𝒕
_ 𝑳𝒂 𝒔 𝑹𝒂 𝑱𝒎 𝒔

𝑣 𝑡

𝑲𝒃

Figure 1: Model of a DC Motor

Parameters La , Kb and Jm are unknown and need to be estimated.


Kt
Jm La
(a) Show that the transfer function from va (t) to ω(t) is as follows: G( s) = a s+ Kt Kb
s2 + R
La Jm La
−
(b) By assuming La = 0.005, Jm = 10−7 and Kb = 5.5 ∗ 10 2 , simulate the system in Simulink.
To do this, you need to draw a block diagram as it is shown in Figure 1. Then apply a
step input to va (t) and plot the input and output during 1 sec. Record va (t) and ω(t)
with the sample time of Ts = 0.01 sec.
(c) Assuming that the discrete transfer function is as G(z) = z2 +a2bz+a1 , write a Matlab code
to find the parameters b, a2 and a1 using Least Squares method.
(d) Show that the assumed discrete transfer function in part (b) is a reasonable assumption.
For example why second order and why no zero.
(e) Using the function “d2c(Gz,’matched’)” in Matlab, find the equivalent continuous transfer
function for G(z) and name it Ĝ(s).
(f) By comparing Ĝ(s) and G(s), find the unknown variables La , Kb and Jm and show that
they are close enough to the parameters you assumed for simulation.
(g) The above simulations showed that the parameters of a DC motor can be estimated when
a step input is applied. However, as you have probably seen in other courses, when the
input is step, there are other methods as well to find the parameters. For example, you
can measure the rise time, peak time and the damped frequency of the response, and then
estimate the poles of the transfer function.
In order to show the advantage of the least squares over the other method, we will estimate
the parameters when the input is not step. This can represent the case when the motor is
connected to a system and working. In this case the input of the motor is determined by
the controller and it is not step anymore. To simulate this, in Figure 1 apply a noise signal
(instead of step signal) and show that still the unknown parameters can be estimated.

1
2. In this problem you are supposed to estimate the moment of inertia of the motor using recursive
least squares (RLS) algorithm. Suppose that the only unknown parameter is Jm and the other
−
parameters are given (La = 0.005 and Kb = 5.5 ∗ 10 2 ). In this simulation we assume that the
inertia is changing (this is realistic as in reality the load attached to the motor can be varying).
For simulation, suppose that the value of Jm is changing sinusoidal between 0.8 ∗ 10−7 and
1.2 ∗ 10−7 with the frequency of 1 Hz. Figure 2 shows how to simulate this case in simulink.
Run the simulation for 5 sec and estimate Jm online using the RLS method. Investigate the
effect of forgetting factor λ on the convergence rate.

Figure 2: Simulating of varying inertia

the equations of RLS algorithm are given below:


  −1 
1
 Pl = λ Pl−1 − Pl−1 al λ + aTl Pl−1 al aTl Pl−1

K l = P l al (1)
 
θ̂l = θ̂l−1 + K l yl − aTl θ̂l−1

Hint: Notice that in this problem there is only one unknown parameter (Jm ) therefore the
dimension of θ is 1x1. What you can do is to use the forward difference (FD) approximation
of s as s = z−1
Ts
and write a difference equation corresponding to Fig. 1. Then keep all known
terms in one side of the equation and take the terms containing θ to the other side.
In this simulation you can (for example) apply a sine wave with amplitude=1 and frequency=
1 rad/sec to the input (va (t)). As an example, Fig. 3 shows the estimated inertia (Jˆm ) where
ˆ
the initial parameters are selected as: Jm(0) = 10−7 , P0 = 1000 and λ = 0.7.

2
Figure 3: Estimated Inertia

You might also like