“Comparison of Nonlinearity Measures based on Time Series Analysis for Nonlinearity Detection”Authors: Seyed Mehrdad Hosseini, Tor Arne Johansen and Alireza FatehiAffiliation: Toosi University of Technology (Iran) and Department of Engineering Cybernetics, Norwegian University of Science and Technology Reference: 2011, Vol. 32, No. 4, pp. 123-140. |
|
Keywords: System Identification, Nonlinearity Measure, Higher Order Spectra, Volterra Series
Abstract: The main purpose of this paper is a study of the efficiency of different nonlinearity detection methods based on time-series data from a dynamic process as a part of system identification. A very useful concept in measuring the nonlinearity is the definition of a suitable index to measure any deviation from linearity. To analyze the properties of such an index, the observed time series is assumed to be the output of Volterra series driven by a Gaussian input. After reviewing these methods, some modifications and new indices are proposed, and a benchmark simulation study is made. Correlation analysis, harmonic analysis and higher order spectrum analysis are selected methods to be investigated in our simulations. Each method has been validated with its own advantages and disadvantages.
PDF (1031 Kb)
DOI: 10.4173/mic.2011.4.1
Currently no DOI forward links to this article
References:
[1] Ashley, R., Patterson, D., and Hinich, M. A diagnostic test for nonlinear serial dependence in time series fitting errors. Journal of Time Series Analysis, 1986. 7:165--178. doi:10.1111/j.1467-9892.1986.tb00500.x.
[2] Barnett, A. and Wolff, R. A time-domain test for some types of nonlinearity. IEEE Transactions on Signal Processing, 2005. 53:26--33. doi:10.1109/TSP.2004.838942.
[3] Bedrosian, E. and Rice, S. The output properties of volterra systems (nonlinear systems with memory) driven by harmonic and gaussian inputs. In Proc. of the IEEE, volume59. pages 1688--1707, 1971. doi:10.1109/PROC.1971.8525.
[4] Bendat, J. and Piersol, A. Engineering applications of correlation and spectral analysis. John Wiley, 1980.
[5] Billings, S. and Chen, S. Identification of nonlinear systems - a survey. IEE Proceeding, 1980. 127:272--285. doi:10.1049/ip-d:19800047.
[6] Billings, S. and Fadzil, M. The practical identification of system with nonlinearities. In IFAC Identification and System Parameter Estimation. pages 117--130, 1985.
[7] Billings, S. and Voon, W. Structure detection and model validity tests in the identification of nonlinear systems. IEE Proceeding, 1983. 130:193--199. doi:10.1049/ip-d:19830034.
[8] Billings, S. and Voon, W. Correlation based model validity tests for nonlinear models. International Journal of Control, 1986. 44:235--244. doi:10.1080/00207178608933593.
[9] Brillinger, D. An introduction to polyspectra. Annals of Mathematical Statistics, 1965. 36:1351--1374. doi:10.1214/aoms/1177699896.
[10] Caillec, J. and Garello, R. Comparison of statistical indices using third order statistics for nonlinearity detection. Journal of Signal Processing, 2004. 84:499--525. doi:10.1016/j.sigpro.2003.11.013.
[11] Choudhury, M., Shah, S., and Thornhill, N. Diagnosis of process nonlinearities and valve stiction. Springer, 2008.
[12] Collis, W., White, P., and Hammond, J. Higher order spectra: The bispectrum and trispectrum. Mechanical Systems and Signal Processing, 1998. 12:375--394. doi:10.1006/mssp.1997.0145.
[13] Evans, C., Rees, D., and Jones, D. Identifying linear models of systems suffering nonlinear distortions, with a gas turbine application. In IEE Proc. Control Theory and Application, volume 142. pages 229--240, 1995. doi:10.1049/ip-cta:19951841.
[14] Evans, C., Rees, D., Jones, D., and Weiss, M. Probing signals for measuring nonlinear volterra kernels. In IEEE Proc. of Instrumentation and Measurement Technology Conf. pages 10--15, 1995. doi:10.1109/IMTC.1995.515094.
[15] Evans, C., Rees, D., and Jones, L. Nonlinear disturbance errors in system identification using multisine test signals. IEEE Transactions on Instrumentation and Measurements, 1994. 43:238--244. doi:10.1109/19.293427.
[16] Fackrell, J. Bispectral analysis of speech signals. Ph.D. thesis, University of Edinburgh, 1997.
[17] Haber, R. Parametric identification of nonlinear dynamic systems based on correlation functions. In IFAC 5th Identification and System Parameter Estimation. 1979.
[18] Haber, R. Nonlinearity test for dynamic process. In IFAC Identification and system Parameter Estimation. 1985.
[19] Hinich, M. Testing for gaussianity and linearity of a stationary time series. Journal of Time Series Analysis, 1982. 3:169--176. doi:10.1111/j.1467-9892.1982.tb00339.x.
[20] Hjellvik, V. and Tjřstheim, D. Nonparametric tests of linearity for time series. Biometrika, 1995. 82:351--368. doi:10.1093/biomet/82.2.351.
[21] Isserlis, L. On a formula for the product-moment coefficient of any order of a normal frequency distribution in any number of variables. Biometrika, 1918. 12:134--139. doi:10.1093/biomet/12.1-2.134.
[22] Kreyzsig, E. Introductory functional analysis with application. John Wiley, 1978.
[23] Ljung, L. System identification-theory for the user. Prentice-Hall, 2nd ed., 1999.
[24] Marzocca, P., Nicholas, J., Milanese, A., Seaver, M., and Trickey, S. Second-order spectra for quadratic nonlinear systems by volterra function series: analytical description and numerical simulation. Mechanical Systems and Signal Processing, 2008. 22:1882--1895. doi:10.1016/j.ymssp.2008.02.002.
[25] McCormack, A., Godfrey, K., and Flower, J. The detection of and compensation for nonlinear effects using periodic input signals. In Intl. Conf. on Control. pages 297--302, 1994. doi:10.1049/cp:19940148.
[26] Nicholas, J., Oslon, C., Michalowicz, J., and Bucholtz, F. The bispectrum and bicoherence for quadratically nonlinear systems subject to non-gaussian inputs. IEEE Transaction of Signal Processing, 2009. 57:3879--3890. doi:10.1109/TSP.2009.2024267.
[27] Nijmeijer, H. and Schaft, A. Nonlinear dynamical control systems. Springer, 1990.
[28] Nuttal, A. Theory and application of the separable class of random processes. MIT Technical Report, 1958.
[29] Pintelon, R. and Schoukens, J. System identification - a frequency approach. Wiley-IEEE Press, 2001.
[30] Priestley, M. Non-linear and non-stationary time series analysis. Academic Press, 1988.
[31] Rao, T. and Gabr, M. A test for linearity of time series analysis. International Journal of Time Series, 1980. 1:145--158.
[32] Rusticelli, E., Ashley, R., Dagum, E., and Patterson, D. A new bisepectral tests for nonlinear serial dependence. Econometric Reviews, 2008. 28:279--293. doi:10.1080/07474930802388090.
[33] Saikkonen, P. and Luukkonen, R. Lagrange multiplier tests for testing non-linearities in time series models. Scandinavian Journal of Statistics, 1988. 15:55--68.
[34] Schetzen, M. The volterra and wiener theories of nonlinear systems. John Wiley, 1980.
[35] Solomou, M. and Rees, D. Frequency domain analysis distortions on linear frf measurements. In Instrumentation and Measurement Technology Conf., volume2. pages 1653--1658, 2003. doi:10.1109/IMTC.2003.1208030.
[36] Szücs, B., Monos, E., and Cz\'aki, F. New aspects of blood pressure control. In IFAC 6th Triannial World Congress. pages 1--10, 1975.
[37] Varlaki, P., Terdik, G., and Lototsky, V. Test for linearity and bilinearity of dynamic systems. In IFAC 7th Symposium on System Identification and Parameter Estimation. pages 427--432, 1985.
[38] Worden, K. and Tomlinson, G. Nonlinearity in structural dynamics: detection, identification and modelling. Inst. of Physics Publ. Bristol and Philadelphia (IOP), 2001.
BibTeX:
@article{MIC-2011-4-1,
title={{Comparison of Nonlinearity Measures based on Time Series Analysis for Nonlinearity Detection}},
author={S. M. Hosseini and T. A. Johansen and A. Fatehi},
journal={Modeling, Identification and Control},
volume={32},
number={4},
pages={123--140},
year={2011},
doi={10.4173/mic.2011.4.1},
publisher={Norwegian Society of Automatic Control}
};
