Документ взят из кэша поисковой машины. Адрес оригинального документа : http://zebu.uoregon.edu/~uochep/talks/talks02/G020254.pdf Дата изменения: Thu Sep 26 00:48:37 2002 Дата индексирования: Tue Oct 2 10:36:15 2012 Кодировка: Поисковые слова: релятивистское двицеойе

LIGO-G020254-00-Z

________________________
Three Veto Schemes
R. Schofield and R. Rahkola (Oregon)

1) Measured Coupling 2) Outlier Rejection 3) Excess Coincidence

________________________


________________________
absGlitch
· Monitor by R. Rahkola - glitchMon skeleton · High- or band-pass filtering · Glitch begins when count threshold is exceeded · Glitch ends when no samples exceed threshold for selectable delay time · Glitch size is the maximum count level reached during glitch · Trigger optional · Outputs rms and kurtosis · Outputs a histogram of glitches; selectable binning

________________________


________________________
1) Measured Coupling
Thresholds from PEM to AS_Q Coupling-Measurements
the night before the run, that were large enough to show up on AS_Q at several times the background rms for the effected frequency band.

1) Generated PEM signals (e.g. banged trash cans)

2) Threshold calculated for PEM channel was, in counts:
threshold = PEMbang_rms (AS_Qb.g._rms / AS_Qbang_rms)

3) Highest threshold selected if multiple thresholds for multiple bands. 4) Ran absGlitch with these thresholds;
using a 30 - 900 Hz bandpass and 0.25 s dead time on 694270001 to 694290000

Channel H0:PEM-PSL2_MIC H0:PEM-PSL2_ACCX H0:PEM-BSC1_MAG1X H0:PEM-BSC1_MAG1Y H0:PEM-BSC1_MAG1Z TOTAL

Threshold (counts) 1000 200 15000 15000 15000

Events 71 64 0 0 0

Fraction of time vetoed 18/20000 16/20000 0 0 0 34/20000

________________________


________________________
Veto candidate: microphone burst
One of the 10 largest mic. candidates, 694270001 - 694290000
Time series
4000

2000

0

Signal

-2000

-4000
H2:IOO-MC_F[0][0]

-6000 0

H0:PEM-PSL2_MIC[0][0]

0.2

0.4

0.6

0.8

1

Time (s)
T0=05/01/2002 16:35:26 Avg=1

Not visible on AS_Q or MC_F time series; possibly visible in MC_F power spectra
Power spectrum
10
3

H2:IOO-MC_F H0:PEM-PSL2_MIC

10

2

Magnitude

10

1

10

-1

1 T0=05/01/2002 16:35:26

10

10

2

Frequency (Hz)
Avg=1 BW=1.5

________________________


________________________
Possibly unwanted veto: single bin glitch

Time series
2000 1000 0

Signal

-1000 -2000 -3000 -4000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

H2:IOO-MC_F[0][0] H0:PEM-PSL2_MIC[0][0] H0:PEM-PSL2_ACCX[0][0]

0.4

0.45

0.5

Time (s)
T0=05/01/2002 17:22:23 Avg=1

8 of 10 largest microphone glitches were single bin (during 694270001 - 694290000) 10 of 10 largest accelerometer glitches were single bin absGlitch outputs glitch length so single bin vetos could be excluded; histograms "single" bin glitches separately

________________________


________________________
2) Outlier rejection
Setting thresholds for outlier vetoes 1) RMS run of absGlitch. Used 1 hour of LHO data to obtain rms and kurtosis for 218 channels 2) AWK routine prepared configuration files with histogram bin widths calculated from rms and kurtosis. 3) Histogram run of absGlitch. Only locked periods were used for H1 and H2. 1 Hz high-pass and 0 dead time on 694270001
694290000

to

4) AWK routine used histograms to prepare configuration files with thresholds set to reject outliers using an approximate interquartile range method: threshold = 7 * level below which the absolute value of 1/2 of the data points lie. 5) "Veto" run of absGlitch. Total time "vetoed" out of 20000 seconds for all "working" PEM channels (112), except BSC9_MAG and TILT: less than 362.25 s ; 2% We have also used a similar scheme for time quota vetoes

________________________


________________________
Channels producing highest veto rates
Channel H0:PEM-BSC9_MAGZ H0:PEM-BSC9_MAGX H0:PEM-COIL_MAGZ H0:PEM-MY_TILTY H0:PEM-EY_TILTY H0:PEM-EX_TILTY H0:PEM-EX_TILTX H0:PEM-MY_TILTX H0:PEM-EY_SEISZ H0:PEM-EY_SEISY H0:PEM-LVEA_TILTX H0:PEM-MX_TILTX H0:PEM-EY_SEISX H0:PEM-LVEA_TILTX H0:PEM-MX_TILTX H0:PEM-EY_SEISX H0:PEM-BSC10_ACC2Z H0:PEM-BSC5_ACCZ H0:PEM-MX_TILTY H0:PEM-BSC10_ACC1Z H0:PEM-BSC5_MIC Threshold (counts) 415.811 768.194 19031.4 11.0548 11.0337 13.8524 14.8922 12.105 193.68 169.786 16.3148 15.9372 182.763 16.3148 15.9372 182.763 277.904 364.907 18.0079 301.122 7327.53 Events 47729 38228 3589 3010 2078 2241 2222 2013 318 250 787 697 187 787 697 187 79 68 316 75 46 Vetoed time (s) out of 20000 9433.7 7370.051 175.809 99.2383 75.6211 68.8164 67.375 59.543 30.0117 26.1367 22.1797 20.6953 16.9062 22.1797 20.6953 16.9062 15.5454 11.8887 11.75 11.0898 6.08887


________________________
3) Excess Coincidence
Thresholds from coincidence rate differences between aligned and misaligned time series Using intersite environmental burst correlations program, a modification of glitchMon

MICROPHONE to MC_F;

1s misalignment; coincidence window: 0.25s, only 1 event allowed per second; total time: 46410 s; H0:PEMPSL2_MIC; H2:IOO-MC_F Threshold (sigma) >= 4.25 3.975 to 4.25 3.7 to 3.975 3.425 to 3.7 3.15 to 3.425 4 3 47 362 2213 Events 0 6 53 368 2226 Offset time series events (on - off) 4 -3 -6 -6 -13 2 3 10 27.0 66.6 sqrt(on+off)

Set the veto level at 4.25 sigma for a total of 4 vetoed seconds in the playground set

________________________


________________________
Excess coincidence may help set MICH burst vetoes
MICH to AS_Q;
Threshold (sigma) >= 10 9.15 to 10 8.3 to 9.15 7.45 to 8.3 6.6 to 7.45 5.75 to 6.6 4.9 to 5.75 4.05 to 4.9 3.2 to 4.05 window: 0.1s, only 1 event allowed per second; H2:LSC-

AS_Q and H2:LSC-MICH_CTRL; Total time 46410 s Events 657 88 81 110 143 189 226 363 1711 Offset time series events 97 25 21 29 45 42 73 155 1335 (on - off) 560 63 60 81 98 147 153 208 376 27.5 10.6 10.1 11.8 13.7 15.2 17.3 22.8 43.0 sqrt(on+off)

Set MICH veto threshold at level where (on - off) coincidence rate during bursts deviates from (on - off) rate when MICH is burst free; use absolute instead of sigma thresholds. This scheme may require us to fix the MICH bursts in order to get a burst-free run..... probably a good idea anyway.

________________________