Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: A fluid level measurement system to accurately determine fluid levels in dynamic environments has been
Received 13 August 2009 described. The measurement system is based on a single ultrasonic sensor and Support Vector Machine
Received in revised form 19 March 2010 (SVM) based signal processing and classification scheme. For exemplification of the measurement system
Accepted 11 May 2010
in dynamic environments, the novel measurement system is experimented and verified on a fuel tank of
Available online 1 June 2010
a running vehicle. The effects of slosh and temperature variations on the acoustic sensor based measure-
ment system are reduced using the novel approach. The novel approach is based on -SVM classification
Keywords:
method with the Radial Basis Function (RBF) to compensate for the measurement error induced by the
Intelligent level measurement
Liquid slosh
sloshing effects in the tank due to the motion of the moving vehicle. In this approach, raw sensor sig-
Radial Basis Function nals are differentiated after smoothing with some selected pre-processing filters, namely, Moving Mean,
Support Vector Machine Moving Median, and Wavelet filter. The derivative signal is then transformed into Frequency Domain to
reduce the size of input features before performing the signal classification with SVM. Field trials were
performed on actual vehicle under normal driving conditions at various fuel volumes ranging from 5 L to
50 L to acquire sample data from the ultrasonic sensor for the training of SVM model. Further drive trials
were conducted to obtain data to verify the SVM results. A comparison of the accuracy of the predicted
fluid level obtained using SVM and the pre-processing filters is provided. It is demonstrated that the -
SVM model using the RBF kernel function and the Moving Median filter has produced the most accurate
outcome compared with the other signal filtration methods in terms of fluid level measurement.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction mechanical level sensors need to occupy a large space and suffer
from the frictional wear-out over a period of time [1]. The impor-
Modern automotive vehicles are equipped with digital gauges tance of level sensor reliability in hostile environments over long
as well as with additional functionalities that inform drivers about periods of time has lead to the introduction of various forms of
their vehicle’s fuel consumption and the remaining distance that motionless level sensors. Ultrasonic devices can be used in con-
the vehicle can travel without refueling. However, the high preci- tainers with pressures up to 2 mega Pascal (MPa), temperatures up
sion digital displays and these additional utilities have to rely on the to 100 ◦ C, and depths up to 30 m, with an accuracy of approximately
accuracy of the level sensor itself. The reliability and accuracy of the 2% [2]. The ultrasonic sensor is one such example of a compact as
fluid level measurement system in a dynamic environment, which well as contact-less proximity sensor that is being investigated to
primarily depends on the level sensor, is increasingly becoming a determine the fluid level in automotive fuel tanks. The ultrasonic
concern for automotive industries as well as the everyday vehicle sensor determines the fluid level by transmitting echo pulses and
user. measuring the return time of the reflected echoes. If the speed of
Conventional fluid level measurement systems determine the sound in the medium is known then the fluid level can be calculated
fluid level with the use of float that is linked with a variable resistor using the following equation:
whose resistance is a function of the fluid level. These conventional
1
level = levelref − v· (1)
2
where levelref is the height of the tank, v is the speed of the sound
∗ Corresponding author. Tel.: +61 3 9239 2148; fax: +61 3 9551 8764.
and is the time-of-flight of the ultrasonic echo (Fig. 1).
E-mail addresses: jenny.terzic@delphi.com (J. Terzic), rnagarajah@swin.edu.au
(C.R. Nagarajah), muhammad.alamgir@delphi.com (M. Alamgir).
However, the speed of sound is influenced by the temperature of
1
Tel.: +61 3 9214 8530. the medium through which it travels [3]. Therefore, changes in the
2
Tel.: +61 4 2301 7656. ambient temperature will create incorrect fluid level readings. The
0924-4247/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.sna.2010.05.005
J. Terzic et al. / Sensors and Actuators A 161 (2010) 278–287 279
Fig. 2. Fuel level signal observed by the level sensor and the calculated average signal in a sample drive trial.
280 J. Terzic et al. / Sensors and Actuators A 161 (2010) 278–287
Fig. 5. Mapping of nonlinearly separable data sets into higher dimensional feature
space.
1
N
1
L(w, , b, , ˛, ˇ, ı) = ||w||2 − + i
2 N
i=1
N
Fig. 6. Relationship between the (a) observed level sensor signal and (b) vehicle
− (˛i (yi (w · xi + b) − + i ) + ˇi i − ı)
acceleration.
i=1
(13)
where L0 is the tank level at still, and f is the unknown sloshing
Some variants of the -SVM have been described in Refs. function that depends on the acceleration effects exhibited on the
[34–36]. The description of the SVM classification and optimization tank, the existing fluid level, and the tank geometry. The goal is
provided so far was related to linearly separable or non-separable focused on determining the actual level L0 using the sensor output
data sets only. In many real-life applications, the data set may L(t) and the function f. The output of the level sensor was observed
not be linearly separable, especial due to the effects of measure- (Fig. 6) to have direct relationship with the vehicle acceleration,
ment errors and noise [37]. To accurately classify data sets that when the vehicle was driven under normal driving conditions.
are nonlinear, the training data sets are mapped onto higher (or After observing the relationship between the acceleration of the
infinite) dimensional space called feature space (Hilbert Space or vehicle and the output signal L(t) from the level sensor, the effects
Inner Product Space). The intuition of mapping data sets into higher of slosh can be minimized by predicting the slosh wave once the
dimensional space is that data can be clearly separated and hence sloshing function f has been determined. Assuming the amount of
better classified in higher dimensional space as illustrated in Fig. 5. fluid in the storage tank to be constant, the actual fuel level L0 in
SVM optimization for the data sets that are not linearly separa- the storage tank under the effects of sloshing can be given as:
ble can be carried out simply by mapping each feature vector x with
L(t)
a mapping function such that x → (x) and then carrying out the L0 = = constant (18)
f
SVM optimization described earlier [38]. For a set of training sam-
ples T = {(x1 ,y1 ), (x2 ,y2 ), . . ., (xn ,yn ) }, where x is an d-dimensional The unknown function f is solved by experimentation with the
feature vector, the mapping of the training data set T is expressed aid of the SVM based approach. An SVM model was constructed
in the feature space as: and trained with the actual driving data obtained through several
field trials. Fig. 7 demonstrates the method adopted to develop an
(T ) = {((x1 ), y1 ), ((x2 ), y2 ), . . . , ((xn ), yn )} (14) accurate fluid level measurement system.
Hence, the decision function D(xk ) for the prediction of the out- For the ultrasonic transducer mounted at a height levelref on top
put can be given as [37]: of the tank, the instantaneous output of the ultrasonic level sensor
n at time t and temperature T can be calculated as:
D(xk )=sign[w · (xk ) + b] = sign ˛i yi (xi ) · (xk ) + b (15) (t)
Level(t, T ) = levelref − v(T ) (19)
2
i=1
where (t) is the time-of-flight at instant t of the ultrasonic echo,
A famous Radial Basis Function (RBF) Kernel function can be and v(T ) is the speed of ultrasonic echo at temperature T. The
expressed as [39]: expression v(T ) can be obtained using Eq. (2).
(xi , xj ) = exp(−||xi − xj ||2 ), >0 (16)
Fig. 11. Typical speed and observed during the experiments for each field trials.
Fig. 9 illustrates the signal output when the Moving Mean and
Fig. 10. Wavelet filter applied on the raw signal. Moving Median filters are applied to the raw signal data. The filtered
versions of the raw signal do not contain high-frequency noise.
The third filter investigated in the experiment is the Wavelet
2.3. Signal enhancement Transform (WT) filter that analyses signals at different frequency
bands by de-composing them into coarse information and detailed
The raw sensor signals were smoothened or filtered to remove information sets. The coarse information set contains the low-
high-frequency noise and spikes before processing the sensor sig- frequency components, and the detailed information set contains
nals through the SVM based signal processing method described the high-frequency components of the input signal. Only low-
previously. The Moving Mean and Moving Median filters slide frequency components, which reflect a smoothened version of the
across the raw signal and calculate the mean/median values in raw signal, are used and the high-frequency components of the raw
the neighbouring sampled points. If x is the sampled raw signal of signal, which usually contain noise, are eliminated. Fig. 10 shows
length N, and w is size of the moving window of the filter, then the the high frequency signal (b) and the low-pass filtered signal (c)
filtered output y using mean and median can be obtained using Eqs. when the raw sensor signal (a) was processed with the Discrete
(20) and (21), respectively. The width of the moving window w is Wavelet Transform (DWT) function. The Wavelet Transformation
set to 20. Therefore, the sliding window takes 20 sampled values of was processed through the MATLAB Wavelet Toolbox [40] using
the raw signal and produces a mean or median value at the output: the dwt function with Daubechies [41] Wavelet (db1).
y[i] = mean(x[i − 1], x[i − 2], . . . , x[i − w]), for w ≤ i ≤ N
(20) 3. Experimental setup
y[i] = mean(x[1], x[2], . . . , x[i]), for 1 ≤ i < w
y[i] = median(x[i − 1], x[i − 2], . . . , x[i − w]), w≤i≤N A fuel tank was fitted with an ultrasonic sensor near the top
(21) center of the tank. The tank can be approximated as a rectangular
y[i] = median(x[1], x[2], . . . , x[i]), for 1 ≤ i < w cuboid with dimensions 34 cm × 34 cm × 81 cm. The fuel tank was
Fig. 12. Block diagram of the SVM system for training and verification of the experimental data.
284 J. Terzic et al. / Sensors and Actuators A 161 (2010) 278–287
Table 1
List of tank volumes investigated in the experiment.
Table 2
Techinical details of the ultrasonic transducer used in the experiment.
Accuracy ±0.32 cm
Output 0.5–4.5 V (min–max)
Resolution 0.18 cm
Operating Temperature −40 to 80 ◦ C
Designed for gasoline and diesel liquids
Table 3
Training parameters used by the LIBSVM train tool.
Fig. 14. Graph of the frequency coefficients obtained from the training data.
J. Terzic et al. / Sensors and Actuators A 161 (2010) 278–287 285
Fig. 15. Validation result generated after training and validating the SVM models for each investigated drive trial.
from the first field trials was used to train the SVM model, and half scaled between an optimum range (0–1) using the LIBSVM scale
of the samples obtained from the second field trials were used for tool.
verification of the network performance. Table 3 lists the training parameters used to train the SVM mod-
The training and validation process is shown in Fig. 13, which els using the LIBSVM train tool. These parameters were selected
was carried out using MATLAB and LIBSVM [42] software appli- experimentally using the LIBSVM tool until a higher classification
cations. LIBSVM [42] is an integrated software for support vector rate was achieved.
classification, regression and distribution estimation. It supports Fig. 14 shows the frequency coefficients of the unfiltered sig-
multi-class classification, which is required for training the ultra- nals obtained using the MATLAB built-in fft function. Trunk [43] has
sonic signals at multiple volume levels. The signal features were demonstrated that data can be detrimental to classification, espe-
286 J. Terzic et al. / Sensors and Actuators A 161 (2010) 278–287
Table 4
Relative average error using statistical averaging methods and SVM based system.
Abs. average error 7.0% (1.53 L) 5.6% (1.33 L) 2.2% (0.57 L) 2.4% (0.63 L) 1.3% (0.39 L) 2.2% (0.59 L)
Max. error 40.6% (5.77 L) 35.4% (5.19 L) 16.1% (3.21 L) 16.1% (3.21 L) 4.8% (1.79 L) 16.1% (3.21 L)
for use in a professional car racing where vehicle is subjected to [23] Y.H. Hu, J.-N. Hwang, Handbook of Neural Network Signal Processing, CRC Press,
a highly dynamic maneuvers. Drivers of cars equipped with this Boca Raton, 2002.
[24] V. Kecman, Learning and Soft Computing: Support Vector Machines, Neural
measurement method can confidently drive higher number of laps Networks, and Fuzzy Logic Models, MIT Press, Cambridge, MA, 2001.
without fear of running out of fuel in situations where fuel level in [25] J. Zhang, X. Liu, J. Liu, F. Peng, J. Tian, Y. Wang, W. Zhang, M. Xie, SVM-based ultra-
the tank is low. sonic medicine image diagnosis, Medical Image Acquisition and Processing
4549 (2001) 92–95.
[26] A.R. Ahmad, C. Viard-Gaudin, M. Khalid, E. Poisson, Online handwriting recog-
6. Future work nition using support vector machine, in: IEEE Region 10 Conference 2004,
2004.
[27] K.M. Rajpoot, N.M. Rajpoot, Wavelets and support vector machines for texture
An ultrasonic sensor coupled with the Support Vector Machine classification, in: Proceedings of International Multitopic Conference (INMIC),
(SVM) approach to signal processing will be used to address other 2004, pp. 328–333.
influencing factors such as atmospheric pressure and the tilt that [28] J. Huang, X. Shao, H. Wechsler, Face pose discrimination using support vector
machines (SVM), in: International Conference on Pattern Recognition, vol. 1,
causes liquid to shift to one side. With the rapid improvements 1998, pp. 154–156.
in microprocessor technology, it will be possible to automatically [29] A. Boni, L. Gasparini, R. Pianegiani, D. Petri, Low-power and low-cost implemen-
train the SVM model in real time, which will further increase tation of SVMs for smart sensors, in: IEEE Instrumentationand Measurement
Technology Conference Proceedings, vol. 1, 2005, pp. 603–607.
the effectiveness of the measurement system in dynamic environ-
[30] R.L. Allen, D.W. Mills, Time-domain signal analysis, in: Signal Analysis: Time,
ments. Frequency, Scale, and Structure, IEEE Press, Wiley–Interscience, Piscataway, NJ,
2004, p. 322.
[31] B. Schölkopf, A. Smola, R. Williamson, P. Bartlett, New support vector algo-
References
rithms, Neural Computation 12 (5) (2000) 1207–1245.
[32] P.-H. Chen, C.-J. Lin, B. Schölkopf, A tutorial on v-support vector machines,
[1] A.C. Fischer-Cripps, Force, pressure and flow, in: Newnes Interfacing Compan- Applied Stochastic Models in Business and Industry 21 (2) (2005) 111–136.
ion, Newnes, Oxford, Boston, 2002, pp. 54–70. [33] S. Abe, Two-class support vector machines, in: Support Vector Machines for
[2] W.C. Dunn, Level measurement, in: Introduction to Instrumentation, Sensors Pattern Classification, Springer, London, 2005, pp. 15–82.
and Process Control, Artech House, Boston, 2005, pp. 115–126. [34] C.-C. Chang, C.-J. Lin, Training v-support vector classifiers: theory and algo-
[3] H. Kuttruff, Ultrasonics—Fundamentals and Applications, Elsevier Applied Sci- rithms, Neural Computation 13 (9) (2001) 2119–2147.
ence, 1991. [35] M.A. Davenport, The 2nu-SVM: A Cost-Sensitive Extension of the
[4] R.A. Serway, J.W. Jewett, Sound waves, in: Physics for Scientists and Engineers, nu-SVM, Journal [serial on the Internet], 2005 Date, Available from:
6th ed., Thomson-Brooks/Cole, Belmont, CA, 2004, pp. 512–542. http://dspace.rice.edu/handle/1911/19831.
[5] Z.-Y. Song, C.-Y. Liu, X.-L. Song, Application research of information fusion tech- [36] A new nu-support vector machine for training sets with duplicate samples, in:
nology of multi-sensor in level measurement, in: International Conference on Y.-S. Jia, C.-Y. Jia, H.-W Qi (Eds.), 2005 International Conference on Machine
Machine Learning and Cybernetics, vol. 6, 2004, pp. 3511–3514. Learning and Cybernetics, Guangzhou, 18–21 August, 2005.
[6] Z. Song, C. Liu, X. Song, Y. Zhao, J. Wang, A virtual level temperature compen- [37] O. Ivanciuc, Applications of support vector machines in chemistry, Reviews in
sation system based on information fusion technology, in: IEEE International Computational Chemistry 23 (2007) 291–400.
Conference on Robotics and Biomimetics, 2007, pp. 1529–1533. [38] B. SchNlkopf, A.J. Smola, Learning with Kernels: Support Vector Machines, Reg-
[7] D.C. Gazis, W.F. Kane, R.J. von Gutfeld, inventors, International Business ularization, Optimization, and Beyond, MIT Press, Cambridge, MA, 2002.
Machines Corporation, assignee, Ultrasonic liquid level gauge for tanks subject [39] C.-W. Hsu, C.-C. Chang, C.-J. Lin, A Practical Guide to Support Vector Classifica-
to movement and vibration, Patent No. 5,793,705, 1996. tion, 2008.
[8] C.M. Combs, P.H. Goodwin Jr., inventors, Robertshaw Controls Company, [40] M. Misiti, Y. Misiti, G. Oppenheim, J.-M. Poggi, Wavelet Toolbox 4—Users Guide,
assignee, Adjustable ultrasonic level measurement device, Patent No. 4221004, MathWorks, 2009.
1978. [41] I. Daubechies (Ed.), Ten Lectures on Wavelets, Society for Industrial and Applied
[9] R.A. Ibrahim, Liquid Sloshing Dynamics: Theory and Applications, Cambridge Mathematics, Philadelphia, PA, 1992.
University Press, Cambridge, UK, New York, 2005. [42] C.-C. Chang, C.-J. Lin, LIBSVM: a library for support vector machines. Software
[10] K. Modaressi-Tehrani, S. Rakheja, R. Sedaghati, Analysis of the overturning available at http://www.csie.ntu.edu.tw/∼cjlin/libsvm, 2001.
moment caused by transient liquid slosh inside a partly filled moving tank, [43] G.V. Trunk, A problem of dimensionality: a simple example, Pattern Analysis
Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Auto- and Machine Intelligence PAMI-1 (3) (1979) 306–307.
mobile Engineering 220 (3) (2006) 289–301. [44] O. Bousquet, U. Luxburg, G. Rätsch, Machine learning summer school, in: O.
[11] L. Dai, L. Xu, A numerical scheme for dynamic liquid sloshing in horizontal Bousquet, U. von Luxburg, G. Rätsch (Eds.), Advanced Lectures on Machine
cylindrical containers, Proceedings of the Institution of Mechanical Engineers, Learning: ML Summer Schools 2003, Canberra, Australia, February 2–14, 2003,
Part D: Journal of Automobile Engineering 220 (7) (2006) 901–918. Tübingen, Germany, August 4–16, 2003, revised lectures, Springer, Berlin, New
[12] S. Wiesche, Computational slosh dynamics: theory and industrial application, York, 2004.
Computational Mechanics 30 (5) (2003) 374–387. [45] F. van der Heijden, R.P.W. Duin, D. de Ridder, D.M.J. Tax, Feature extraction
[13] N.C. Sinha, S.K. Pal, R.K. Bhattacharyya, Experimental investigation of slosh and selection, in: Classification, Parameter Estimation, and State Estimation:
dynamics of liquid-filled containers, Experimental Mechanics 41 (2001) 63–69. An Engineering Approach using MATLAB, Wiley, Chichester, West Sussex, Eng.,
[14] H.-S. Kim, Y.-S. Lee, Optimization design technique for reduction of sloshing Hoboken, NJ, 2004, pp. 183–14.
by evolutionary methods, Journal of Mechanical Science and Technology 22
(2008) 25–33.
[15] D. Reed, J. Yu, H. Yeh, S. Gardarsson, Investigation of tuned liquid dampers Biographies
under large amplitude excitation, Journal of Engineering Mechanics 124 (4)
(1998) 405–413.
[16] H. Kobayashi, H. Obayashi, inventors, Nissan Motor Company, Limited, Jenny Terzic Quality/Six Sigma Manager at Delphi Automotive Systems Australia
assignee, Fuel volume measuring system for automotive vehicle. Patent No. – Powertrain Division. Holds Bachelor of Mechanical Engineering (with honours)
4611287, 1983. and Master of Engineering in Computer Integrated Manufacturing. Has extensive
[17] H. Kobayashi, T. Kita, inventors, Nissan Motor Company, Limited assignee, Fuel experience in vehicle fuel system design, development and testing. Area of research
gauge for an automotive vehicle, Patent No. 4470296, 1982. include: neural networks, support vector machines, intelligent sensors and applica-
[18] T. Guertler, M. Hartmann, K. Land, A. Weinschenk, inventors, DAIMLER BENZ AG tions of artificial intelligence in automotive vehicle engineering.
(DE) assignee, Process for determining a liquid quantity, particularly an engine
oil quantity in a motor vehicle, Patent No. 5831154, 1997. Romesh Nagarajah Professor of Mechanical Engineering at Swinburne University
[19] S. Zhao, Remote sensing data fusion using support vector machine, in: IEEE of Technology. Graduated with an Honours degree in Mechanical Engineering from
International Geoscience and Remote Sensing Symposium Proceedings, vol. 4, the University of Ceylon in 1973, M.Phil. in Robotics from the University of Notting-
2004, pp. 2575–2578. ham in 1985 and a Ph.D. from Swinburne University of Technology in 1994. He has
[20] R.A. Reyna, D. Esteve, D. Houzet, M.-F. Albenge, Implementation of the SVM neu- received several national competitive grants, named in several international patents
ral network generalization function for image processing, in: IEEE International and published 75 papers in international journals and conferences.
Workshop on Computer Architectures for Machine Perception (CAMP’00),
Muhammad Alamgir Electrical Engineer at Delphi Automotive Systems Australia.
2000.
Graduated in Computer Systems Engineering from RMIT University in 2008. Areas
[21] V.N. Vapnik, Statistical Learning Theory, Wiley, New York, 1998.
of research include smart sensors, artificial neural networks and support vector
[22] V.N. Vapnik, The Nature of Statistical Learning Theory, Springer, New York,
machines.
London, 1995.