9512.net

# Signal Processing Algorithms_图文

Signal Processing Algorithms
Ivy Xu S3211523

Signal Processing Algorithms Design Overview
Actuators: 29 in total LH: Good 10 Samples LH: Bad 9 Samples RH: Bad 10 Samples Data Inputs 29 Subjective Evaluation ratings Psychoacoustic Metrics Objectives Use the data of these 29 actuators to design a Signal Processing Algorithms which has the capability to identify good or bad on the manufacturing assembly line. Methods Gate Noise Diagnosis Design Linear Regression and Least Squares Method Tools Excel and MATLAB FFT

Historical Information on Noise Problem
The noises which denote a bad actuator is not only limited to a certain Sound Pressure Level (dB level), but may include: intermittent clicking sharp pitch changes significant differences between motor driving forward and backward.

Psychoacoustic Metrics

Seven Parameters
Sound Pressure Levels—decibel (dB) Loudness—Sone Roughness—Asper Sharpness—Acum Fluctuation Strength Tonality Articulation Index--%

Gate Noise Diagnosis Design Considerations
SPL Loudness Roughness Sharpness Fluc. Strength Tonality AI

Range of each parameter:
42.5 < SPL< 49.6 dB 3.99<Loudness<6.29 soneGD 0<Roughness<0.03 Asper

dB(A)

SonesGD

Asper

Acum

Vacil

Tu

%

Samples =
Sample 1 47.90 48.70 48.90 48.40 49.10 49.40 45.20 46.00 46.00 47.80 47.60 47.70 47.00 46.20 46.20 48.80 47.80 48.00 47.30 48.20 48.30 48.70 49.50 49.60 47.10 46.20 46.50 47.80 47.70 47.50 47.70 49.60 45.20

LH_0 ---> 'GOOD'
5.56 5.76 5.84 5.59 5.81 5.90 4.64 4.91 4.93 5.62 5.36 5.44 5.18 4.83 4.82 5.55 5.18 5.28 5.27 5.48 5.39 6.02 6.16 6.29 4.96 4.73 4.91 5.36 5.44 5.37 5.39 6.29 4.64 0.00489 0.00054 0.00378 0.00000 0.00021 0.00000 0.00046 0.00000 0.00006 0.00436 0.00000 0.00000 0.00599 0.00217 0.00041 0.00139 0.00000 0.00000 0.02860 0.01600 0.01200 0.00041 0.00000 0.00002 0.00000 0.00000 0.00000 0.00077 0.00070 0.00407 0.00 0.03 0.00 2.86 2.81 2.74 2.79 2.78 2.79 2.73 2.72 2.75 2.79 2.85 2.81 2.69 2.78 2.76 2.96 2.99 2.96 2.77 2.77 2.88 2.77 2.80 2.70 2.77 2.73 2.69 2.90 2.96 2.92 2.81 2.99 2.69 0.0380 0.0365 0.0367 0.0421 0.0380 0.0386 0.0375 0.0366 0.0371 0.0354 0.0395 0.0408 0.0341 0.0405 0.0353 0.0386 0.0367 0.0410 0.0337 0.0295 0.0287 0.0334 0.0319 0.0376 0.0421 0.0398 0.0353 0.0406 0.0367 0.0377 0.04 0.04 0.03 0.103 0.156 0.182 0.112 0.167 0.193 0.097 0.127 0.108 0.150 0.155 0.158 0.108 0.089 0.106 0.163 0.132 0.137 0.067 0.081 0.085 0.194 0.175 0.199 0.136 0.132 0.139 0.146 0.117 0.107 0.13 0.20 0.07 92.40 92.10 92.40 92.60 92.00 92.00 95.50 94.90 94.90 93.40 93.70 93.80 94.30 95.00 95.10 92.20 93.40 93.10 92.90 92.00 91.90 92.70 91.60 91.90 94.00 94.90 94.70 93.30 92.90 93.00 93.29 95.50 91.60

Sample 2

Sample 3

Sample 4

Sample 5

2.57<Sharpness<2.99 Acum 0.03<Fluctuation Strength<0.04 Vacil 0.05<Tonality<0.20 Tu 91.6<Articulation Index<95.5

Sample 6

Sample 7

Sample 8

Sample 9

Sample 10

Average Max Min

Gate Noise Diagnosis Design Considerations cont.
Actuators
Reject Gate1-SPL
42.5 < SPL< 49.6 dB

Reject
Gate3-Roughness 0<R<0.03 Asper

Reject
Gate4-Sharpness 2.57<S<2.99 Acum

Gate2-Loudness 3.99<L<6.29 Sone

Pass

Pass

Pass Pass

Pass

GOOD Actuators
Not True!

Gate7-Articulation Index 91.6<AI<95.5

Pass
Gate6-Tonality 0.05<T<0.20

Pass

Reject
Gate5-Fluctuation Strength 0.03<FS<0.04

Reject

Reject

Reject

SPL Loudness Roughness Sharpness Fluc. Strength Tonality AI

dB(A)

SonesGD

Asper

Acum

Vacil

Tu

%

Samples =
Sample 1 44.80 44.90 45.50 46.80 46.60 46.40 43.30 43.00 42.50 47.80 49.00 48.90 47.50 47.50 47.70 54.20 53.50 53.50 50.40 50.60 50.50 49.00 49.40 49.40 43.70 42.50 42.50 47.46 54.20 42.50

LH_x ---> 'FAILED'
4.24 4.28 4.49 5.05 5.04 5.02 4.15 4.16 4.06 5.50 5.95 5.91 5.36 5.32 5.37 7.81 7.72 7.63 6.51 6.58 6.53 6.12 6.00 6.09 4.12 3.99 4.00 5.44 7.81 3.99 0.00998 0.01290 0.00729 0.00122 0.00043 0.00121 0.00128 0.00000 0.00016 0.00018 0.00063 0.00081 0.00583 0.00206 0.00249 0.15000 0.14400 0.14200 0.00078 0.00394 0.00644 0.00024 0.00062 0.00587 0.00000 0.00151 0.00031 0.02 0.15 0.00 2.51 2.49 2.49 2.56 2.57 2.57 2.60 2.62 2.57 2.57 2.50 2.53 2.64 2.65 2.64 3.08 3.01 3.08 2.72 2.75 2.76 2.59 2.51 2.49 2.64 2.61 2.63 2.64 3.08 2.49 0.0342 0.0333 0.0294 0.0366 0.0321 0.0302 0.0400 0.0381 0.0349 0.0368 0.0323 0.0333 0.0353 0.0338 0.0320 0.0314 0.0378 0.0415 0.0405 0.0317 0.0350 0.0415 0.0322 0.0350 0.0369 0.0376 0.0372 0.04 0.04 0.03 0.078 0.084 0.095 0.138 0.142 0.139 0.122 0.120 0.117 0.133 0.204 0.195 0.125 0.152 0.162 0.239 0.225 0.224 0.165 0.159 0.154 0.171 0.253 0.266 0.123 0.087 0.113 0.15 0.27 0.08 97.40 97.40 97.00 94.60 94.50 94.50 97.20 97.10 97.50 93.80 92.80 93.00 93.60 93.50 93.50 82.60 83.50 83.60 88.90 88.20 88.20 92.60 93.10 93.10 97.30 97.90 98.00 93.13 98.00 82.60

Sample 2

Sample 3

Sample 4

Three Gates only to reject all LH Bad ones. LH actuators are more sensitive to SPL, Loudness and Sharpness

Sample 5

Sample 6

Sample 7

Sample 8

Sample 9

Average Max Min

RH_x_1 Rejection
RH_x_1_2m ( 0.00- 3.07 s).FFT vs t (4096,66.6%,HAN). f/Hz 20k 10k 5k

RH_x_1 Reject
? Roughness and

2k

1k

500

Pitch Change ? Change from fold-in to fold-out
Tonality AI

200

100

50 Right

RH_x_1
0.5 0 10 1 20 t/s L/dB[SPL] 2 40 2.5 50 3

20

60

SPL

Loudness

Roughness

Sharpness

Fluc. Strength

dB(A)

SonesGD

Asper

Acum

Vacil

Tu

%

Samples =
RH_x_1 45.90 46.10 45.50 43.90 44.20 43.70 49.40 49.90 49.90 46.70 46.30 45.90 48.10 48.10 48.40 46.50 46.70 46.70 45.90 46.20 45.70

RH_x ---> 'FAILED'
4.62 4.74 4.63 4.18 4.25 4.10 5.53 5.66 5.53 4.97 4.83 4.76 5.44 5.47 5.45 4.76 4.75 4.81 4.79 4.88 4.84 0.00002 0.00132 0.00000 0.00012 0.00000 0.00000 0.00026 0.00071 0.00039 0.00000 0.00000 0.00802 0.00044 0.00138 0.00110 0.00002 0.00099 0.01220 0.00000 0.00000 0.00140 2.88 2.81 2.81 3.03 3.11 3.10 2.99 2.93 2.97 3.00 3.02 2.99 3.22 3.28 3.31 2.92 2.88 2.86 3.09 3.10 3.08 0.0348 0.0329 0.0356 0.0373 0.0401 0.0422 0.0310 0.0295 0.0264 0.0323 0.0314 0.0298 0.0297 0.0315 0.0266 0.0350 0.0334 0.0329 0.0326 0.0322 0.0355 0.084 0.134 0.115 0.057 0.061 0.059 0.082 0.105 0.120 0.077 0.069 0.087 0.080 0.057 0.066 0.045 0.054 0.081 0.090 0.101 0.115 93.90 94.00 95.00 95.60 94.90 95.50 89.50 88.80 89.00 92.50 92.90 93.40 90.90 90.90 90.80 92.20 91.90 92.00 93.90 94.00 93.80

RH_x_2

RH_x_3

RH_x_4

RH_x_5

RH_x_7

RH_x_8

Other Considerations on Gate Method
The ranges for each parameter is probably not precise and accurate. How to choose the appropriate gates ? What extra gates should be considered? Objective measurement only

Subjective Rating 10 0 1 2 3 4 5 6 7 8 9

LH_0_1 LH_0_2 LH_0_3 LH_0_4 LH_0_5 LH_0_6 LH_0_7 LH_0_8 LH_0_9 LH_0_10 LH_X_1 LH_X_2 LH_X_3 LH_X_4 LH_X_5 LH_X_6 LH_X_7 LH_X_8 LH_X_9 RH_X_1 RH_X_2 RH_X_3 RH_X_4 RH_X_5 RH_X_6 RH_X_7 RH_X_8 RH_X_9 RH_X_10
Group Result Trained Evaluator Samples

Comparison of Subjective Evaluation
Comparison of Untrained and Trained Evaluation

Linear Regression and Least Squares Method
Linear Regression
Linear regression is a form of regression analysis in which the relationship between independent variables and dependent variable is modelled by a least square function.

Least Squares Method
Least squares corresponds to the maximum likelihood criterion and can be derived as a method of moment estimator.

Estimators
Least squares method and regression analysis are conceptually different; however, the method of least squares is often used to generate estimators in regression analysis.

Relationship between Subjective Evaluation and Psychoacoustic Metrics

Mathematical Equations
Equation 1: Subjective Rating
SR=A0+A1*L+A2*R+A3*FS+A4*S+A5*T+A6*SPL+A7*AI

Equation 2: Subjective Rating Estimation
SR/SR0=A0+A1*(L/L0)+A2*(R/R0)+A3*(FS/FS0)+A4*(S/S0) +A5*(T/T0)+A6*(SPL/SPL0) +A7*(AI/AI0)

LH0_10

9

6.77

0.00102

0.0213

2.84

0.233

50.30

87.50

Linear Regression for all 29 samples
SR differences for all 29
5.00 SR differences of all 29 4.00

A0

0
3.00

A1

0.1256
2.00

A2 A3 equal to

-0.002 0.9616
differences

1.00

0.00 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87

A4

-0.1332
-1.00

A5

-0.0181
-2.00

A6

-0.2312
-3.00

A7

0.0024

-4.00

-5.00 No. of testing

Linear Regression for 19 LHS samples
SR differences for LH 19
A0
3.00

0

A1
2.00

0.6618

1.00

A2 A3 equal to

-0.0037 0.0612

Differences

0.00 1 -1.00 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57

A4

4.6792

-2.00

A5

-0.0064

-3.00

A6

-4.4494

-4.00
No. of testing

A7

0.0236

SR L R FS S T SPL AI = 0.6618 ? 0.0037 + 0.0612 + 4.6792 ? 0.0064 ? 4.4494 + 0.0236 SR0 L0 R0 FS0 S0 T0 SPL0 AI 0

Linear Regression for 10 RHS samples
10 RHS plus LH0_10

A0

0

1.00

A1

0

A2
0.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

0.0068 equal to -0.0083

A3

-1.00

A4

0.0092

-2.00

A5

-0.0101

-3.00

A6

0.4367

A7
-4.00

0.0122

SR R FS S T SPL AI = 0.0068 ? 0.0083 + 0.0092 ? 0.0101 + 0.4367 + 0.0122 SR0 R0 FS 0 S0 T0 SPL0 AI 0

Noise Characteristics of LH and RH Actuators
LH: A1, A4, A6 have big absolute values LH actuators are more sensitive to Loudness, Sharpness and Sound Pressure Levels (SPL). RH: A5, A6, A7 have big absolute values RH actuators are more sensitive to Tonality, SPL and Articulation Index (AI).

New Noise Diagnosis Gate Design
Separate LH and RH Combination of Subjective Evaluation and Objective Method (Psychoacoustic Metrics).
Several Gates LH Actuators
Gate1-SPL
42.5 < SPL< 49.6 dB

Good LH Actuators

Reject Several Gates RH Actuators
Gate1-SPL
42.5 < SPL< 49.6 dB

Good RH Actuators Reject

Thanks for Attention!
Questions?

Signal Processing Algorithms_图文.ppt
Signal Processing Algorithms_工学_高等教育_教育专区。Signal Processing Algorithms Ivy Xu S3211523 Signal Processing Algorithms Design Overview Actuator ...
Indexing terms Algorithms, Signal processing_图文.pdf
Indexing terms Algorithms, Signal processing_专业资料 暂无评价|0人阅读|0次下载 | 举报文档 Indexing terms Algorithms, Signal processing_专业资料。digital ...
3_Signal Processing_图文.ppt
Introduction of Weak GPS Signal Processing ? ? Software GPS Receiver Filtering Algorithms GPS Signal Characteristics Satellite Signal Waveforms GPS Signal ...
EEG signal processing 脑电信号处理方法算法_图文.ppt
EEG signal processing 脑电信号处理方法算法_生物学...Classification algorithms Common methods COMMON ...
Introduction to Digital Signal Processing_图文.ppt
A simple digital processing system Digital signal processing algorithms ( [ '?lg?ri??m ] 算法) ...
Adaptive Signal Processing5_教学案例/设计_教学研究_教育专区。Adaptive Signal ...Basic of other adaptive algorithms ? ? A recursive method to yield solution...
...ALGORITHMS FOR DIGITAL SIGNAL PROCESSING AGC_图....pdf

Adaptive Signal Processing1_教学案例/设计_教学研究_教育专区。Adaptive Signal ...algorithms should offers a tracking capability for a non-stationary signal ...
signal-processing-toolbox_图文.pdf
Signal Processing Toolbox 6.15 Perform signal processing, analysis, and algorithm development Signal Processing Toolbox? provides industry-standard algorithms for...
Adaptive Signal Processing3_教学案例/设计_教学研究_教育专区。Adaptive Signal ...above This is the start point of many other algorithms Non-parametric ...
Adaptive Signal Processing10_教学案例/设计_教学研究_教育专区。Adaptive Signal ... Learning algorithms Learning with a teacher/Learning without a teacher...
Approximate signal processing using incremental ref....pdf
Approximate signal processing using incremental refinement and deadlinebased algorithms - A frame...
BATTERY- FRIENDLY DESIGN OF SIGNAL PROCESSING ALGORITHMS.pdf

Electrocardiogram signal processing method for exac....doc

SIGNAL PROCESSING TOOLSET.pdf

Shanghai Jiaotong University SEIEE Adaptive Signal Processing 6. Basic Adaptive Algorithms Wang Kaizhi from Institute of Signal Processing Method of Steepest ...
digital single processing_图文.ppt
, is a particularly suitable for digital signal processing microprocessor, its main application is to realize various digital signal processing algorithms. ...
Interpolation of Non-Linear Retinex Type Algorithms_图文.pdf
Interpolation of Non-Linear Retinex Type Algorithms_专业资料。image processing,...signal processing, where new samples are interpolated between lowresolution ...
Signal Processing in Mobile Wireless Communications_图文_....ppt
Signal Processing in Mobile Wireless Communications_工学_高等教育_教育专区。...What is SP Input signals of interests Desired outputs Techniques Algorithms ...
...A family of normalized LMS algorithms_图文.pdf