Beruflich Dokumente
Kultur Dokumente
Raman spectral analysis integrated with multivariate calibration is a fast which results in the intensity of the cosmic spikes being smaller
and effective solution to monitor chemical product properties. However, than those in other literature and more difficult to detect.
Raman instruments utilizing charge-coupled device (CCD) detectors Spectral processing should be quick so that the information can
suffer from occasional spikes caused by cosmic rays. Cosmic spikes can
be utilized in process optimization of the gasoline properties
disturb or even destroy the meaningful chemical information expressed by
normal Raman spectra. In online monitoring, some cosmic spikes have
online.
intensity and bandwidth similar to normal Raman peaks of chemical A wealth of approaches have been proposed in the literature
components when a low resolution and cost-effective Raman instrument is for the detection and removal of cosmic spikes. Some of these
used. Moreover, the online Raman spectra always contain variations of approaches can only detect whether a spectrum contains spikes
strong Raman peaks and fluorescence. Current spike-removal methods and cannot find their exact positions in the spectrum.6,7 Others
seem to have difficulty detecting and recovering cosmic spikes in these can both detect and remove spikes; these approaches fall into
online Raman spectra. Therefore, an improved algorithm is proposed. In four categories.
this algorithm, a new scheme composed of intensity identification and The first category can be defined as single-scan methods,
local moving window correlation analysis is introduced for cosmic spike
including robust smoothing filter,8 weighted moving window
detection; intensity identification based on derivative spectra and local
linear fitting approximation are used for the recovery of cosmic spikes.
filter,9 missing point polynomial filter,10,11 and wavelet
The algorithm is proved to be simple and effective and has been applied in recognition.12 These approaches are based on the hypothesis
an online Raman instrument installed at a continuous catalytic reforming that the bandwidth of cosmic spikes is much smaller than the
unit in a refinery. bandwidth of the substance of interest. However, the resolution
Index Headings: Raman spectroscopy; Online monitoring; Cosmic spikes; of our Raman instrument is low and gasoline is a relatively
Charge-coupled device; CCD detector; Data processing. complex mixture, so the bandwidth of some Raman peaks is
similar to that of some cosmic spikes. Some normal peaks will
be mistaken for cosmic spikes if these methods are used.
Moreover, filter-based approaches may distort gasoline Raman
INTRODUCTION spectra.
The second category, including the robust summation
Raman spectral analysis is commonly applied in the method13 and the upper-bound spectrum method14 and its
chemical industry,1–5 as vibration information represented by improved versions,15,16 is based on the fact that the probability
Raman spectral peaks is specific to the chemical bonds and of cosmic spikes appearing at the same CCD pixel in two
symmetry of molecules. It has become a suitable tool for real- sequential spectra is very low and at least one of the same-
time monitoring of chemical product quality because of its high position pixels in two spectra is normal. These methods cost
sampling rate and reliability. Charge-coupled device (CCD) more spectral acquisition time, which limits their application in
detectors are widely used in Raman spectrometers because of online monitoring.
their high sensitivity and low noise, but they are vulnerable to The third category tries to improve optical hardware
cosmic rays, which lead to visible spikes in Raman spectra. performance to avoid spikes, such as image curvature
Typically, these spikes have narrow bandwidth, positive correction,17 which will increase instrumental complexity and
unidirectional peaks, and random positions. Cosmic spikes cost.
pollute or even distort the effective Raman signals of chemical A fourth category of approaches is proposed to detect and
products. remove the cosmic spikes in Raman spectral images.18–20
The aim of our research team is to monitor the properties of These approaches are based on the fact that the probability of a
reformed gasoline in a continuous catalytic reforming unit cosmic spike appearing at the same CCD pixel in two spatially
using an online Raman instrument. Unfortunately, the cosmic adjacent spectra is low and that the difference between the
spikes result in unexpected abrupt fluctuations of gasoline spectrum of a spatial pixel and its neighbors in the spatial
properties, so they must be detected and removed. In our domain is very small. The currently detected spectrum is
application, a low-resolution and cost-effective Raman instru- approximated by the most similar neighboring spectrum. This
ment is used, so some cosmic spikes have similar bandwidth to spectrum can also be called the nearest neighbor of the
some normal Raman peaks; the spectra are obtained over a currently detected spectrum. The spikes are detected by
long integration time because the Raman signal of reforming comparing the currently detected spectrum and its approximat-
gasoline is not strong enough in our spectral apparatus; the ed spectrum. Therefore, these methods are also called the
existing time of cosmic spikes is less than the integration time,
nearest neighbor methods.
In online reformed gasoline monitoring, the change of
Received 22 October 2010; accepted 17 August 2011. Raman spectra is slow and smooth under normal operation. A
* Author to whom correspondence should be sent. E-mail: lkdai@iipc.zju.
edu.cn. series of spectra obtained just before the current spectrum in a
DOI: 10.1366/10-06169 time domain can be taken as the neighboring spectra, and the
0003-7028/11/6511-1300$2.00/0
1300 Volume 65, Number 11, 2011 Ó 2011 Society for Applied Spectroscopy APPLIED SPECTROSCOPY
neighboring spectral information can also be utilized just as the areas. The detection zone around an SRP area is defined as zs,
nearest neighbor methods. However, a more effective spike and the remaining detection zones are defined as zw. Each
detection and removal algorithm is needed because changes in detection zone zw is confirmed to contain a spike and the
strong Raman peaks and fluorescence for online spectra result detection zone zs will be checked in the third step. (If the pixel
in visible difference between the currently detected spectrum number of zs is less than 3, neighboring normal pixels are
and its approximated spectrum. The simple intensity identifi- introduced to insure that the correlation analysis can be
cation used in previous nearest neighbor methods is not realized.)
suitable. Third, the moving window correlation analysis between x
An improved cosmic spike removal algorithm based on a and xa at zs is used to detect spikes on the strong Raman peaks.
new spike identification scheme is proposed for online The window size is denoted as w and the analysis result as Rc.
monitoring of a low-resolution Raman system. In this new Rc is a vector composed of correlation coefficients of each w-
scheme, both intensity identification and local moving window pixel window. zs is determined to contain a spike if the
correlation analysis are used to distinguish spectral variations minimum of Rc is less than Rt. Conversely, zs is judged as the
of strong Raman peaks from real cosmic spikes and confirm normal zone at which the intensity of a strong Raman peak
whether a spike is just a strong Raman peak. Intensity varies. The moving window correlation analysis is based on the
identification based on derivative spectra and local linear fact that the cosmic spikes have more narrow width than the
fitting approximation are used for the recovery of cosmic strong Raman peaks.
spikes. The algorithm is conceptually simple and effective for Cosmic Spike Removal. According to the detection of
online cosmic spike removal. Furthermore, it is very easy to cosmic spikes, the removal procedure is specific to each kind of
implement and costs little calculation time. spike zone. For spike zone zw , a local linear fitting
approximation between x and xa in zw is implemented. The
THEORY neighboring nr normal pixels at two sides of zw are used to
estimate the fitting coefficients. The locally approximated
Strong Raman Peak Area Definition. In order to spectrum in zw is used to replace the contaminated pixels.
distinguish spikes on strong Raman peaks, the strong Raman For a spike zone zs, zs should be resized since zs also
peak (SRP) area is defined in advance. First, automated contains the variation of strong Raman peaks and the size of zs
subtraction of fluorescence based on the iterative polyfit is larger than the real spike zone. Because variation of the
method21 is carried out on the current spectra. Second, we gradient intensity for strong Raman peaks is much smaller than
obtain the maximum of the fluorescence subtracted spectrum that for cosmic spikes, derivative spectra of x and xa are first
im. Third, the partition ratio rp is introduced and the area in calculated and are denoted as xd and xda. Then, the absolute
which the intensity of the fluorescence subtracted spectrum is residue of xd and xda is obtained and its standard deviation is
more than im 3 rp is defined as the SRP area. calculated and defined as sxd. A scaled absolute residual vector
Most Similar Neighboring Spectrum Definition. In our rd is given as follows:
algorithm, the spectrum just before the current spectrum at the
last sampling time can be defined as the most similar jxd xda j
rd ¼ ð4Þ
neighboring (MSN) spectrum since in process monitoring the sxd
sample at the last sampling time has the most similar chemical
composition with the current sample. Third, an intensity threshold tr is introduced and the resized
Linear Approximation of Current Spectrum. Following spike zone zsr in which the intensity of rd is greater than tr is
the spatial nearest neighbor method by Zhang and Henson,14 finally obtained in zs.
linear approximation of the current spectrum is implemented After resizing zs into zsr, the same local linear fitting
for neighboring comparison as follows: approximation is used to recover x at zsr.
Discontinuity of the Detection Zone. The detection zones
x ¼ axm þ b1 þ e ð1Þ may be discontinuous in two cases. Case 1: a spike is on a
strong Raman peak that is decreasing, in which case some
xa ¼ axm þ b1 ð2Þ spike pixels may have lower intensity than t in r. Case 2: the
absolute residual vector rd is used and the intensity at the pixel
where xm denotes the MSN spectrum; x is the current raw which is the center of the spike is always close to zero. When
spectrum; xa is the linear fit approximated spectrum; a and b discontinuity happens, pixels having lower intensity are
are the linear fitting coefficients; and e is the fitting error. ignored in the detection zones.
Cosmic Spike Detection. For a current spectrum, a scaled Algorithm Procedure. The procedure of the improved
absolute residual vector r is given by: algorithm is as follows:
jx xa j (1) Use the current spectrum after fluorescence subtraction and
r¼ ð3Þ the partition ratio rp to acquire the SRP areas.
s
(2) Obtain the approximated spectrum xa according to Eq. 2.
where s is the standard deviation of the linear fit. (3) Calculate the scaled residual vector r using Eq. 3. Obtain
In our new scheme, an intensity threshold t and a correlation detection zones z according to the intensity threshold t and
coefficient threshold Rt are introduced. divide them into zs and zw. Each zw can be taken as a spike
First, the pixels with intensity greater than t in r are zone. Examine each zs using the moving window
identified and the index corresponding to each set of pixels correlation analysis to confirm whether zs is a spike zone.
above t is defined as a detection zone z. (4) For a spike zone zs, zs is resized to zsr by the scaled
Then, the detection zones are checked if they are in the SRP absolute residual vector rd and threshold tr. Finally, a local
SIMULATION EXPERIMENT
Sample Generation. A series of synthesized online Raman
spectra S(k) consist of the following three parts:
SðkÞ ¼ RðkÞ þ BðkÞ þ NðkÞ ð5Þ
where R(k) denotes the pure Raman signal, B(k) is the curved
baselines to represent the fluorescence, N(k) is random noise,
and k is the time index.
The pure Raman signal can be described by R(k) = fRi(k)ji
= 1, . . ., ng:
X
M
2Aj ðkÞcL; j
Ri ðkÞ ¼ ð6Þ
2
j¼1 p½cL; j þ 4ði m0; j Þ2
where n represents the number of pixels; M is the number of FIG. 1. Two simulated online Raman spectra without spikes.
Lorentzian peaks; and m0,j, c0,j, and Aj(k) denote the central
position, the width, and the area of the jth peak, respectively. 8
In this experiment, we set M = 5, n = 2000, k = 1 and 2, >
> Si ð2Þ þ 15000 i ¼ 353
>
>
cL,1 = 30, m0,1 = 200; cL,2 = 30, m0,2 = 500; cL,3 = 15, m0,3 = < Si ð2Þ þ 2000ð5 ji 500jÞ i 2 ½498; 502
1000; cL,4 = 40, m0,4 = 1100; and cL,5 = 10, m0,5 = 1700. For Ŝi ð2Þ ¼ Si ð2Þ þ 2000ð3 ji 1003jÞ i 2 ½1002; 1004
>
>
real online Raman spectra, Raman peak intensity varies >
> S ð2Þ þ 2000ð3 ji 1126jÞ i 2 ½1125; 1127
: i
according to operating conditions, which can be realized by Si ð2Þ else
adjusting the following peak areas in our simulation: ð14Þ
5
A1 ðkÞ ¼ 10 3½1 þ ðk 1Þ 3 0:02 ð7Þ
INDUSTRIAL ONLINE APPLICATION
A2 ðkÞ ¼ 106 3½1 þ ðk 1Þ 3 0:15 ð8Þ
Instrumental Setup. An online Raman instrument is
installed at a catalytic reforming unit in a refinery. As shown
A3 ðkÞ ¼ 106 3½1 ðk 1Þ 3 0:1 ð9Þ
in Fig. 3, the reformed gasoline is sampled at two sides of a
regulating valve V1, which is located at the bottom outlet of a
A4 ðkÞ ¼ 106 ð10Þ distillation column after a heat exchanger. The instrument
includes reformed gasoline sampling equipment, the Raman
A5 ðkÞ ¼ 106 ð11Þ system, and a central control system. In the Raman system, a
FIG. 5. The spike detection procedure. (a) The SRP area on the current
spectrum; (b) the current spectrum x and its approximated spectrum xa; (c) the
FIG. 4. Real online gasoline spectra. absolute residual vector r.
1 353rd pixel zw
2 494th;504th pixels zs
3 979th;1023rd pixels zs
4 1125th;1127th pixels zw
5 1686th;1713rd pixels zs
FIG. 8. The result of the resizing procedure. (a) The absolute residual vector
rd at detection zone zs (the 494th;504th pixels); (b) the absolute residual
vector rd at detection zone zs (the 979th;1023rd pixel).
FIG. 7. The derivative spectrum of the current spectrum and its approximated
spectrum. (a) The derivative spectra xd and xda at detection zone zs (the FIG. 9. The result of the spike removal procedure. (a) The current spectrum x;
494th;504th pixels); (b) the derivative spectra xd and xda at detection zone zs (b) the recovered current spectrum; (c) the residue of the current spectrum and
(the 979th;1023rd pixels). the recovered current spectrum.
Spectrum Spike pixel Spike pixel Spectrum Spike pixel Spike pixel
index index number index index number
difference at two sides of the fourth spike zone, which will lead
to unusual discontinuity in the recovered x.
Industrial Online Application. Spike removal algorithm
parameters are given in Table III. The first spectrum is set to be
the initial MSN spectrum of the second spectrum and it is
confirmed that there is no cosmic spike on it, so 999 online
Raman spectra will be processed. Twenty-nine (29) cosmic
spikes are found in 29 online spectra, respectively. All of the
contaminated spectra are recovered; the spike distribution in
detail can be seen in Table IV. There are no spikes on the
strong Raman peaks. Two of the recovered spectra (the 476th
FIG. 10. The spike detection procedure by the method of Zhang and Henson. (a) and 599th spectra) are shown in Fig. 11.
The current spectrum x and its approximated spectrum xa; (b) the residual vector r1.
Figure 12 shows the detection zone for the 162nd spectrum. 1. P. Marteau, N. Zanier-Szydlowski, A. Aoufi, G. Hotier, and F. Cansell,
Figure 12a illustrates the visible baseline change of the whole Vib. Spectrosc. 9, 101 (1995).
spectrum and the variation of the strong Raman peak at the 2. M. Digne, K. Marchand, and P. Bourges, Oil Gas Sci. Technol. 62, 91
(2007).
410th pixel between the 162nd spectrum and the 161st 3. S. J. Choquette, S. N. Chesler, and D. L. Duewer, Anal. Chem. 68, 3525
spectrum, which indicates that the operating condition of the (1996).
catalytic reforming process fluctuated at that moment. Figure 4. P. E. Flecher, W. T. Welch, S. Albin, and J B. Cooper, Spectrochim. Acta,
12b illustrates that there is strong fluctuation both at the 368th Part A 53, 199 (1997).
pixel and the 410th pixel in the scaled absolute residual vector 5. J. B. Cooper, Chemom. Intell. Lab. Syst. 46, 231 (1999).
r. Finally, two detection zones are obtained, which are the 6. W. J. Egan and S. L. Morgan, Anal. Chem. 70, 2372 (1998).
7. P. J. de Groot, G. J. Postma, W. J. Melssen, L. M. C. Buydens, V. Deckert,
368th pixel and the 406th;413th pixel. The SRP areas are also and R. Zenobi, Anal. Chim. Acta 446, 71 (2001).
obtained in advance, which are the 312th;313th pixel and the 8. B. M. Bussian and W. Harale, Appl. Spectrosc. 38, 309 (1984).
404th;413th pixel. Therefore, the first detection zone is zw 9. Y. Katsumoto and Y. Ozaki, Appl. Spectrosc. 57, 317 (2003).
and is confirmed to be a spike zone. The second detection zone 10. G. R. Phillips and J. M. Harris, Anal. Chem. 62, 2351 (1990).
is zs and the result of the moving window correlation analysis 11. W. Hill and D. Rogalla, Anal. Chem. 64, 2575 (1992).
for zs can be seen in Fig. 13, which shows that the second 12. F. Ehrentreich and L. Summchen, Anal. Chem. 73, 4364 (2001).
13. H. Takeuchi, S. Hashimoto, and I. Harada, Appl. Spectrosc. 47, 129
detection zone contains no spike but a varying strong Raman (1993).
peak. 14. D. M. Zhang, K. N. Jallad, and D. Ben-Amotz, Appl. Spectrosc. 55, 1523
(2001).
CONCLUSION 15. D. M. Zhang and D. Ben-Amotz, Appl. Spectrosc. 56, 91 (2002).
16. D. M. Zhang, J. D. Hanna, and D. Ben-Amotz, Appl. Spectrosc. 57, 1303
An improved algorithm is proposed in this paper for cosmic (2003).
spike removal of Raman spectra obtained by a low-resolution 17. J. Zhao, Appl. Spectrosc. 57, 1368 (2003).
and cost-effective Raman system in online monitoring. The 18. C. J. Behrend, C. P. Tarnowski, and M. D. Morris, Appl. Spectrosc. 56,
1458 (2002).
change of strong Raman peaks in the time domain can be 19. L. Zhang and M. J. Henson, Appl. Spectrosc. 61, 1015 (2007).
distinguished from real cosmic spikes by the algorithm and the 20. U. B. Cappel, I. M. Bell, and L. K. Pickard, Appl. Spectrosc. 64, 195
false detection rate is low. Spikes on both the strong Raman (2010).
peaks and at weak Raman signal areas can all be well 21. C. A. Lieber and A. Mahadevan-Jansen, Appl. Spectrosc. 57, 1363 (2003).