9512.net
甜梦文库
当前位置:首页 >> 工学 >>

Signal Processing Algorithms


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

LH Bad Actuators are rejected
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
2.00 10 RH Bad

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?



更多相关文章:
signal processing methods_图文.ppt
Signal processing is the mathematics, the algorithms and techniques used to manipulate some signal. Often the initial signal is in time domain.Non treated ...
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 ...
A Low-Complexity Time-Domain Signal Processing Algorithm for ....pdf
we propose a time-domain signal processing (TDSP) algorithm aiming at saving the complexity by inserting time-domain smoothing signal wi (n) which ...
【测控专业英语】Signal Process信号处理_图文.ppt
? The high-level purpose of most signal processing is to yield knowledge of a situation so that decisions can be made. 大多数信号处理的高级目标是从...
Signal Processing in....pdf
Signal Processing in..._专业资料。OBJECTIVE Tenure-track teaching and ...B. Giannakis, “Optimal distributed stochastic routing algorithms for wireless ...
Signal Processing Image Communication A software framework ....doc
signal or processing or algorithms or product or development or TI or Freescale) and (multicore or DSP or Development or Tools or Balanced or signal or...
数字媒体课件1(Fundamental Knowledge for Signal Processing)_....ppt
数字媒体课件1(Fundamental Knowledge for Signal Processing)_英语考试_外语学习_教育专区。Fundamental Knowledge for Signal Processing ...
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 ...
...of compressive sensing in ECG and EEG signal processing_....pdf
Implementation of compressive sensing in ECG and EEG signal processing_专业资料。Avibl lea aleoni tanwww cenerc.om sicditce一 lSinerc lbecDi ...
Algebraic signal processing theory Cooley-Tukey typ....pdf
1 Algebraic Signal Processing Theory: Cooley-Tukey Type Algorithms for DCTs and DSTs Markus P¨ schel, Senior Member, IEEE, and Jos? M. F. Moura, ...
Signal processing model and receiver algorithms for a higher ....pdf
Signal processing model and receiver algorithms for a higher rate multiuser TR-UWB communic_专业资料。Transmit-reference (TR) schemes are commonly used only...
Signal processing and communication with solitons_....pdf
Signal processing and communication with solitons Traditional signal processing algorithms rely heavily on models that are inherently linear. Such models are ...
BATTERY- FRIENDLY DESIGN OF SIGNAL PROCESSING ALGORITHMS.pdf
BATTERY- FRIENDLY DESIGN OF SIGNAL PROCESSING ALGORITHMS_专业资料。Portable systems today are designed with lowering the energy consumption as the primary ...
...of Multirate Signal Processing Algorithms_免费下....pdf
Looped Schedules for Dataflow Descriptions of Multirate Signal Processing Algorithms The synchronous dataflow (SDF) programming paradigm has been used ...
Signal Processing Theory and Methods.pdf
Signal Processing Theory and Methods_信息与通信_工程科技_专业资料。[in the ...This can be viewed as an analysis problemhow do simple algorithms ...
Published in J. Signal Processing.pdf
Signal Processing_专业资料。Implementation and experimental evaluation of the ... to voxels for images representation in iterative reconstruction algorithms. ...
Stochastic search for signal processing algorithm ....pdf
Keywords: evolutionary algorithms, signal processing, performance optimization, genetic algorithms 1 Introduction Signal processing algorithms take as an input a ...
...tuning of iterative algorithms in signal processing.pdf
Performance tuning of iterative algorithms in signal processing_专业资料。Presented high-level synthesis describes scheduling for wide class of DSP algorithms. ...
Digital Signal Processing A Performance Comparison.pdf
We think recon gurable systems have special advantages for e ective use in signal processing algorithms because they do not need to fetch, decode and ...
unit 13 Signal Process.doc
While we appear to prefer the total objectivity of a mathematically formulated method of signal processing, it is now well proven that the AI methods ...
更多相关标签:

All rights reserved Powered by 甜梦文库 9512.net

copyright ©right 2010-2021。
甜梦文库内容来自网络,如有侵犯请联系客服。zhit325@126.com|网站地图