Recording the random white noise through the error microphone, Mod1/AI0

Generating transfer coefficients of the system

Transfer coefficients are stored and will be implemented in the ANC Filtered-X LMS adaptive filter

Sound Pressure (Pa) measured from all 16 microphones with respect to time (ms)

Data will determine what frequencies the rocket engine produces and will help produce non-linear shock wave model of noise

Peak sound pressure was 1000 Pa = 153.98 dB

**Active Noise Control for Rocket Engine Noise**

Overview

Future Work

Statement

Rocket Engine produces sound pressure levels in excess of 153dB.

McGregor, TX is 3 miles NE of the test facility.

Goal: Reduce sound pressure by at least 6dB.

Approach

Use the ANC system to invert the incoming noise signal

Play the inverted signal through high intensity speakers

Summary

Single tone frequencies were reduced successfully.

ANC system generates good filter coefficients and models a control waveform that mirrors the rocket noise.

Long Term Goals

Implementation in enviromental settings, Bryan RC Airfield

Implementation of ANC System at SpaceX in McGregor, TX.

Short Term Goals

Programming MIMO system and evaluating results in lab

Dirk de Haan,

Jonathan Jolly,

Ryan Marietta &

Sergio A. Rodriguez

Undergraduate Seniors of Department of Electrical and Computer Engineering

Yong-Joe Kim, Ph.D.

Assistant Professor of Department of Mechanical Engineering

Director of Acoustics and Signal Processing Laboratory

Date: 03/27/2014

Task 1: On-site measurements & data analysis

Measurement setup at Merlin Test site.

Placement of the microphones was to understand noise propagation characteristics in the space.

16 total microphones were used for measurements, 4 high pressure microphones and 12 low pressure microphones.

The high-pressure microphones are appropriate for this high-level noise measurement. However, they are expensive and only four high-level microphones are available. Thus, they are place where the highest noise levels are measured.

Generating an MP3 file from the measured rocket noise data for testing

Experimental setup & Hardware

ANC software development using LabVIEW

LMS algorithm

Filtered-X LMS adaptive filter

Lab performance

Task 2: Development of SISO ANC software and in-lab performance evaluation

Phase I

Task 3: Development of MIMO ANC Software and In-lab Performance Evaluation

Enhance SISO ANC system

SISO ANC testing

Method behind the MIMO ANC software

Task 4: On-site implementation of the MIMO ANC system and performance evaluation

Continue to develop the MIMO system.

To test in the Bryan RC airfield to simulate environmental conditions

Evaluate the results from the field and lab results and make adjustments

Implementation at SpaceX

Phase II

Agenda

Statement/Approach

Research Design and Method

Overview of PHASE I

PHASE II:

TASK 3: Development of MIMO ANC Software and In-lab Performance Evaluation

Future Work: TASK 4: Environmental implementation and performance evaluation.

Conclusion

Questions or Comments

Research Design and Method

Phase 1:

Task 1: On-site noise measurements and data analysis.

Task 2: Development of SISO ANC software and in-lab performance evaluation.

Phase II:

Task 3: Development of MIMO ANC Software and In-lab Performance Evaluation

Task 4: Environmental implementation of the MIMO ANC System and performance evaluation

This graph shows the overall dB levels of the Rocket

The overall average level was 136 dB

Sound pressure measurements collected on-site at SpaceX-McGregor

Onsite Noise Measurement Procedure shown below

Laptop

3 Microphones (b&K 2671)

1 Reference mic

1 Error mic

1 Performance evaluation mic

2 Speakers (BX5a deluxe)

Source & control speakers

NI-DAQ (NI-PXIe-1082)

CompactRIO (NI cRIO-9022)

Module NI 9234 (input)

Module NI 9269 (output)

Hardware

Graphical spectrum of sound pressure with a range frequencies over time

Lower frequencies have greater sound pressure levels compared to higher frequencies

Lower frequencies travel further than higher frequencies over a distance due to absorption.

MIMO ANC setup

LMS Algorithm

ANC Filtered-X LMS Adaptive Filter

LMS Algorithm Coefficients

1 reference mic

2 Control Speakers

2 error mics

Graphical representation of transfer coefficients

The transfer coefficients converge about the x-axis

The ANC Filtered-X LMS Adaptive Filter is ready to be used.

Mod1/AI1 is the reference microphone: x(n)

Simulated rocket noise is recorded through this microphone

Mod1/AI0 is the same error microphone from the LMS algorithm: e(n)

Recording the diminished waveform from the source and anti-noise speakers

Mod2/AO0 is the anti-noise speaker: y(n)

Produces the inverted waveform to cancel the source noise

100Hz Single Tone Test

300Hz Single Tone Test

200Hz Single Tone Test

400Hz Single Tone Test

Average dB before: 135dB

Average dB After: 121dB

Average reduction: 14dB

Percentage drop: 10.37%

Rocket Noise Test

Enhancement of SISO ANC system and setup

Average dB before: 152.5dB

Average dB after: 130dB

Average reduction: 22.5dB

Percentage drop: 17.31%

Average dB before: 147.5dB

Average dB After: 136.5dB

Average reduction: 11dB

Percentage drop: 7.45%

ANC Real Time

LMS Algorithm Continued

Rotated control speaker to face the same direction as the rocket noise simulator.

The combination of the sound waves will sum to the diminished noise.

Integrate an external pulse generator to the system.

LMS Algorithm Equations

π₯(π)=πππππππ π€βππ‘π ππππ π π πππππ

π(π)=π₯(π)βπ(π)

π(π)= White noise input from microphone

ΞΌ=convergence coefficient

(constant)

π(π)=πππππ π πππππ (recorded white noise)

h(n+1)= h(n)+ΞΌ* e(n)*d(n)

w(n+1)=w(n)-Β΅*h(n)*x(n)*e(n)

y(n+1)=y(n)+x(n)*w(n)

h(n+1) = updated transfer coefficient vector

Average dB before: 157dB

Average dB After: 144dB

Average reduction: 13dB

Percentage drop: 8.28%

Graphical representation of the control output y(n), and the reference input noise x(n).

The control output is to inversely match the delayed reference input noise as it is generated from the filtered-X LMS adaptive filter.

Since x(n) is the input into the system, y(n) is the output to generate delayed inverted wave through the control speaker.

The noise level plot is the error microphone.

The secondary path is the transfer function coefficients.

Experimental Conclusion

Rocket engine noise was measured at the SpaceX Merlin Testing Stand with 16 spatially-distributed microphones.

The maximum noise level was 154 dB at 34 meters away from the stand.

It is found that the noise components below 500 Hz contribute significantly to the overall noise level. Thus, these low frequency noise components will be aimed to be reduced with the proposed ANC system.

The ANC system has been built and evaluated in Dr. Kimβs Acoustics lab.

The current ANC system with one error microphone and one control speaker was proven to reduce the simulated rocket engine noise levels by 1-2 dB.

The ANC performance is expected to be further improved by having two error microphones and two control speakers (MIMO) and by focusing the control in the low frequency range below 500 Hz. The current target frequency band is below 2 kHz.

ANC Equations

Active Noise Control

Average dB before: 141.2dB

Average dB after: 140.6dB

Average reduction: 0.6dB

Percentage drop: 0.42%

Block Diagram of ANC System

Questions or Comments?

Results

Pressure vs. Time

Data Collection Setup

Color Spectrum

Overall dB

Microphone Setup