United Arab Emirates University

1
United Arab Emirates University College of
Engineering Graduation Project (II) Course
Design and Implementation of Noise Control System
Using Data Acquisition with MATLAB

• Group members
• Mona Al Abri 200220567
• Khawlah Al Shehhi 200212436
• Moza Nagher 200211359
• Amal Al Manei 200202464

Advisor Dr. Qurban Ali
Second Semester 2007-2008
2
Outline

• Executive Summary
• Objectives
• Background Theory
• MATLAB Simulation
• Hardware components
• Off-line modeling
• Problems
• Conclusion

3
Executive Summary (1/3)

• The project is about designing and implementation
  of active noise control system in a duct using
  data acquisition with MATLAB.
• The basic idea is to cancel the (low frequency)
  unwanted disturbance.

4
Executive Summary (2/3)

• How?
• By generating an anti-phase signal.

5
Executive Summary (3/3)
6
Objectives
7
Noise Problem and Ways of Noise Reduction

• Acoustic noise problems become more and more
  evident as increased numbers of large industrial
  equipment such as
• Engines
• Fans
• Transformers
• compressors

8
Noise Problem and Ways of Noise Reduction (1/2)

• There are two approaches to control acoustic noise

1. Passive Noise Control
Barrier
Noise
2. Active Noise Control

Residual Noise
9
General Applications of ANC Systems (2/2)

• Automotive Noise attenuation inside vehicle
  passenger compartments.
• Appliances Air conditioning ducts,
  refrigerators, washing machines, vacuum cleaners,
  and so on.
• Industrial Fans, air ducts, transformers, power
  generators, blowers, compressors, pumps, wind
  tunnels noisy plants, headphones, and so on.
• Transportation Airplanes, ships, boats,
  helicopters, diesel locomotives, and so on.

10
MATLAB Simulation of FXLMS Algorithm with
Practical Conditions

• Single-tone noise
• Multi-tone noise

11
Simulation for Single-Tone Noise

• Simulation Parameters

Exp. 3 Exp. 2 Exp. 1 Simulation parameters
24 16 8 Size of adaptive filter (M)
0.001 0.001 0.001 Step size (µ)
fo100Hz, fs2000Hz fo100Hz, fs2000Hz fo100Hz, fs2000Hz sinusoidal signal
fo200Hz fo200Hz fo200Hz Single-tone noise
FIR filter FIR filter FIR filter Filter
10 10 10 Averaging loop (A)
1000 1000 1000 of iterations (N)
12
Simulation Results for Single-Tone Noise (1/3)
13
Simulation Results for Single-Tone Noise (2/3)
14
Simulation Results for Single-Tone Noise (3/3)
15
Simulation for Multi-Tone Noise

• Simulation Parameters

Exp. 3 Exp. 2 Exp. 1 Simulation parameters
20 16 8 Size of adaptive filter (M)
0.001 0.001 0.001 Step size (µ)
fo100Hz, fs2000Hz fo100Hz, fs2000Hz fo100Hz, fs2000Hz sinusoidal signal
fo170Hz, fo2200Hz fo170Hz, fo2200Hz fo170Hz, fo2200Hz Multi-tone noise
FIR filter FIR filter FIR filter Filter
10 10 10 Averaging loop (A)
1000 1000 1000 of iterations (N)
16
Simulation Results for Multi-Tone Noise (1/3)
17
Simulation Results for Multi-Tone Noise (2/3)
18
Simulation Results for Multi-Tone Noise (3/3)
19
Hardware Components (1/7)

• PVC pipe
• Length 168 cm.
• Diameter 16 cm.

20
Hardware Components (2/7)

• Two reference microphones.
• Two error microphones.
• Frequency response 30Hz-20KHz
• Connecting wire 2 x 4000mm
• Operating Voltage 1.5 3V

21
Hardware Components (3/7)

• 2 loudspeakers
• 1 noise source loudspeaker
• 1 cancelling loudspeaker
• Frequency Response 45Hz to 20KHz.
• Diameter 16 cm

22
Hardware Components (4/7)

• Power amplifier with two channels.

23
Hardware Components (5/7)

• Fan (noise source)

24
Hardware Components (6/7)

• Function Generator
• To produce a signal with low frequency
• Output the signal through the loudspeaker

25
Hardware Components (7/7)

• Pentium-4 desktop PC with full version of MATLAB,
  and including Data Acquisition and Signal
  Processing Tool Boxes

26
Prototype
27
Programs and Software used in ANC System

• MATLAB with data acquisition tool box (version
  1.7 or above).
• Provides a complete set of tools for analog
  input, analog output, and digital I/O from the
  data acquisition card.
• DAQ Adapter
• allows MATLAB users direct access to analog and
  digital I/O data

28
Off-Line Modeling Technique (1/2)

• FXLMS algorithm requires knowledge of the
  transfer function of the
• Primary path P(z)
• Secondary path S(z)
• Feedback path F(z)
• Off-line modeling can be used to estimate these
  paths during an initial training stage.
• At the end of the training interval, the
  estimated model is fixed and used for ANC
  operation.

29
Off-Line Modeling Technique (2/2)

• The off-line modeling procedure is summarized as
  follows
• Generate random noise signal.
• Obtain desired signal from a sensor (microphone).
• Apply adaptive filter algorithm to get the FIR
  model.

30
Off-Line Modeling
P(z)
S(z)
F(z)
31
System Identification (1/3)

• 1. Primary Path P(z)
• Generate a random noise at the noise source
  point, and record the signal at the location of
  error microphone.
• Take this input-output data to MATLAB, and use
  LMS to get an appropriate order FIR model for the
  primary path P(z).

32
System Identification (2/3)

• 2. Secondary Path S(z)
• Generate a random noise at the location of
  cancelling loudspeaker.
• Record the signal at the location of error
  microphone.
• Take this input-output data to MATLAB, and
  use LMS to get an appropriate order FIR model for
  the  S(z).

33
System Identification (3/3)

• 3. Feedback Path F(z)
• Generate a random noise at the location of
  cancelling loudspeaker.
• Record the signal at the location of reference
  microphone (noise source point).
• Take this input-output data to MATLAB, and
  use LMS to get an appropriate order FIR model for
  the  F(z).

34
MATLAB Code for recording the Reference signal
and the Desired Response
35
Code for Off-line Modeling
36
Problems

• DAQ
• Data acquisition tool box
• FIR model for the paths
• Time limitation

37
Conclusion

• The ANC system was studied and analyzed.
• LMS and FXLMS was implemented by MATLAB.
• The materials of the prototype were selected and
  bought.
• The prototype was built.