Hidden Markov model parameters estimation with independent multiple observations and inequality constraints
- Author(s): Lisha Xia 1 ; Huajing Fang 1 ; Xiaoyong Liu 1
-
-
View affiliations
-
Affiliations:
1:
School of Automation, Huazhong University of Science and Technology, Wuhan 430074, People's Republic of China
-
Affiliations:
1:
School of Automation, Huazhong University of Science and Technology, Wuhan 430074, People's Republic of China
- Source:
Volume 8, Issue 9,
December 2014,
p.
938 – 949
DOI: 10.1049/iet-spr.2013.0505 , Print ISSN 1751-9675, Online ISSN 1751-9683
In this study, the authors focus on hidden Markov model (HMM) parameters estimation with independent multiple observations and non-linear inequality constraints. The parameters estimation process is divided into four steps: initialisation, parameters pre-estimation, parameters re-estimation and termination. The pre-estimation results are used to approximate non-linear inequality constraints to linear inequality constraints. In parameters re-estimation step, the active-set optimisation is combined with the expectation maximisation (EM) algorithm in M-step and the active set-based EM algorithm is proposed to re-estimate HMM parameters when inequality constraints are not satisfied in pre-estimation. An auxiliary function is devised for reconstructing the optimisation objective function and the convergence of the proposed algorithm is also demonstrated. Simulation results indicate that the proposed algorithm provides better performance by modifying the random error of observation data appropriately and it is powerful for industry process fault diagnosis.
Inspec keywords: parameter estimation; approximation theory; fault diagnosis; hidden Markov models; expectation-maximisation algorithm; nonlinear estimation; optimisation
Other keywords: approximate nonlinear inequality constraint; HMM parameter estimation; initialisation step; random error modification; optimisation objective function reconstruction; termination step; hidden Markov model parameter estimation; parameters reestimation step; independent multiple observation; expectation maximisation algorithm; active set-based EM algorithm; industry process fault diagnosis; active-set optimisation; parameters pre-estimation step; M-step algorithm; linear inequality constraint
Subjects: Interpolation and function approximation (numerical analysis); Optimisation; Markov processes; Optimisation techniques; Numerical analysis; Optimisation techniques; Numerical approximation and analysis; Probability theory, stochastic processes, and statistics; Interpolation and function approximation (numerical analysis); Markov processes; Statistics
References
-
-
1)
-
12. Park, S., Choi, S.: ‘A constrained sequential EM algorithm for speech enhancement’, Neural Netw., 2008, 21, (9), pp. 1401–1409 (doi: 10.1016/j.neunet.2008.03.001).
-
-
2)
-
14. Martin, M., Daniel, N., Lenka, L.: ‘Constraints in particle swarm optimization of hidden Markov models’. Intelligent Data Engineering and Automated Learning – IDEAL 2006, 2006, vol. 4224, pp. 1399–1406.
-
-
3)
-
4. Gunter, S., Bunke, H.: ‘HMM-based handwritten word recognition: on the optimization of the number of states, training iterations and Gaussian components’, Pattern Recognit., 2004, 37, (10), pp. 2069–2079 (doi: 10.1016/j.patcog.2004.04.006).
-
-
4)
-
8. Huda, S., Yearwood, J., Togneri, R., et al: ‘A stochastic version of expectation maximization algorithm for better estimation of hidden Markov model’, Pattern Recognit. Lett., 2009, 30, (14), pp. 1301–1309 (doi: 10.1016/j.patrec.2009.06.006).
-
-
5)
- M.M. Aleix , J. Vitria . Learning mixture models using a genetic version of the EM algorithm. Pattern Recognit. Lett. , 8 , 759 - 769
-
6)
-
25. Fletcher, R.: ‘Practical methods of optimization’ (John Wiley & Sons, New York, 1981), part 2.
-
-
7)
-
3. Pecht, M., Jaai, R.: ‘A prognostics and health management roadmap for information and electronics-rich systems’, Microelectron. Reliab., 2010, 50, (3), pp. 317–323 (doi: 10.1016/j.microrel.2010.01.006).
-
-
8)
-
22. Chen, Z.S., Yang, Y.M.: ‘Fault diagnostics of helicopter gearboxes based on multi-sensor mixtured hidden Markov models’, J. Vib. Acoust., 2012, 134, (3), pp. 0310101–0310108.
-
-
9)
- S.X. Ding , P. Zhang , A. Naik , E.L. Ding , B. Huang . Subspace method aided data-driven design of fault detection and isolation systems. J. Process Control , 1496 - 1510
-
10)
-
2. Levinson, S.E., Rabiner, L.R., Sondhi, M.M.: ‘An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition’, Bell Syst. Tech. J., 1983, 62, (4), pp. 1035–1074 (doi: 10.1002/j.1538-7305.1983.tb03114.x).
-
-
11)
-
7. Votsi, I., Limnios, N., Tsaklidis, G.E., et al: ‘Hidden Markov models revealing the stress field underlying the earthquake generation’, Phys. A, 2013, 392, (13), pp. 2868–2885 (doi: 10.1016/j.physa.2012.12.043).
-
-
12)
-
1. Rabiner, L.R.: ‘A tutorial on hidden Markov models and selected applications in speech recognition’. Proc. IEEE, 1989, vol. 77, no. 2, pp. 257–286.
-
-
13)
-
15. Hu, X.H., Eberhart, E.: ‘Solving constrained nonlinear optimization problems with particle swarm optimization’. Proc. of the Sixth World Multiconference on Systems, Cybernetics and Informatics, 2002.
-
-
14)
-
13. Otterpohl, J.R., Emmert-Streib, F., Pawelzik, K.: ‘A constrained HMM-based approach to the estimation of perceptual switching dynamics in pigeons’, Neurocomputing, 2001, 38–40, pp. 1495–1501 (doi: 10.1016/S0925-2312(01)00511-2).
-
-
15)
-
32. Li, Z.F., Fang, H.J., Xia, L.S.: ‘Increasing mapping based hidden Markov model for dynamic process monitoring and diagnosis’, Expert Syst. Appl., 2013, 41, (2), pp. 744–751 (doi: 10.1016/j.eswa.2013.07.098).
-
-
16)
-
19. Huda, M.S., Yearwood, J., Ghosh, R.: ‘A hybrid algorithm for estimation of the parameters of hidden Markov model based acoustic modeling of speech signals using constraint-based genetic algorithm and expectation maximization’. Proc. of Int. Conf. on Computer and Information Science (ICIS), 2007.
-
-
17)
-
5. Parvez, M.T., Mahmoud, S.A.: ‘Arabic handwriting recognition using structural and syntactic pattern attributes’, Pattern Recognit., 2013, 46, (1), pp. 141–154 (doi: 10.1016/j.patcog.2012.07.012).
-
-
18)
-
28. Lee, J., Kang, B., Kang, S.H.: ‘Integrating independent component analysis and local outlier factor for plant-wide process monitoring’, J. Process. Contr., 2011, 21, (7), pp. 1011–1021 (doi: 10.1016/j.jprocont.2011.06.004).
-
-
19)
- S. Huda , J. Yearwood , R. Togneri . A constrained-based evolutionary learning approach to the expectation maximization for optimal estimation0 of the hidden Markov model for speech signal modeling. IEEE Trans. Syst. Man Cybern.t B Cybern. , 1 , 182 - 197
-
20)
-
27. Hsu, C.C., Chen, M.C., Chen, L.S.: ‘A novel process monitoring approach with dynamic independent component analysis’, Control Eng. Pract., 2010, 18, (3), pp. 242–253 (doi: 10.1016/j.conengprac.2009.11.002).
-
-
21)
-
15. Ge, Z.Q., Song, Z.H.: ‘Process monitoring based on independent component analysis-principal component analysis (ICA–PCA) and similarity factors’, Ind. Eng. Chem. Res., 2007, 46, (7), pp. 2054–2063 (doi: 10.1021/ie061083g).
-
-
22)
-
24. Simon, D., Chia, T.L.: ‘Kalman filtering with state equality constraints’, IEEE Trans. Aerosp. Electron. Syst., 2002, 38, (1), pp. 128–136 (doi: 10.1109/7.993234).
-
-
23)
- Z. Michalewicz , M. Schoenauer . Evolutionary algorithms for constrained optimization problems. Evol. Comput. , 1 - 32
-
24)
-
17. Chau, C.W., Kwong, S., Diu, C.K., et al: ‘Optimization of HMM by a genetic algorithm’. Proc. of the IEEE Int. Conf. on Acoustics, Speech, and Signal Processing (ICASSP), 1997, vol. 3, pp. 1727–1730.
-
-
25)
-
6. Elliott, R.J.: ‘Exact adaptive filters for Markov chains observed in Gaussian noise’, Automatica, 1994, 30, (9), pp. 1399–1408 (doi: 10.1016/0005-1098(94)90004-3).
-
-
26)
-
16. Paquet, U., Engelbrecht, A.P.: ‘A new particle swarm optimiser for linearly constrained optimization’. Proc. of the Congress on Evolutionary Computation 2003 (CEC'2003), 2003, vol. 1, pp. 227–233.
-
-
27)
-
9. Dempster, A.P., Laird, N.M., Rubin, D.B.: ‘Maximum likelihood from incomplete data via the EM algorithm’, J. R. Stat. Soc. B, 1977, 39, (1), pp. 1–38.
-
-
28)
-
10. McLachlan, G.J., Basford, K.E.: ‘Mixture models: inference and applications to clustering’ (Dekker, New York, 1988).
-
-
29)
-
30. Chiang, L.H., Russell, E., Braatz, R.D.: ‘Fault detection and diagnosis in industrial systems’ (Springer Verlag, 2001).
-
-
30)
-
21. Liu, P., Liu, C., Jiang, H., et al: ‘A constrained line search optimization method for discriminative training of HMMs’, IEEE T. Audio Speech, 2008, 16, (5), pp. 900–909 (doi: 10.1109/TASL.2008.925882).
-
-
31)
-
23. Chen, Z.S., Yang, Y.M., Hu, Z.: ‘Fault prognosis of complex mechanical systems based on multi-sensor mixture hidden semi-Markov models’, Proc. Inst. Mech. Eng. C, J. Mech. Eng. Sci., 2013, 227, (8), pp. 1853–1863 (doi: 10.1177/0954406212467260).
-
-
32)
-
26. Downs, J.J., Vogel, E.F.: ‘A plant-wide industrial process control problem’, Comput. Chem. Eng., 1993, 17, (3), pp. 245–255 (doi: 10.1016/0098-1354(93)80018-I).
-
-
1)