Identification and Monitoring of Automotive Engines 1

The objective of this paper is to extend and refine the nonlinear canonical variate analysis (NLCVA) methods developed in the previous work for system identification and monitoring of automotive engines. The use of additional refinements in the nonlinear modeling are developed including the use of more general bases of nonlinear functions. One such refinement in the NLCVA system identification is the selection of basis functions using the method of Leaps and Bounds with the Akaike information criterion AIC. Delay estimation procedures are used to considerably reduce the state order of the identified engine models. This also considerably reduces the number of estimated parameters that directly affects the identified model accuracy. This increased accuracy also affects the ability to monitor changes or faults in dynamic engine characteristics. A further objective of this paper is the development and use of nonlinear monitoring methods as extensions of several previously used linear CVA monitoring procedures. For the case of linear Gaussian systems, these monitoring methods have optimal properties in detecting faults or system changes in terms of the general maximum likelihood method. In the nonlinear case, departures from optimality are investigated, but the procedure is shown to still work quite effectively for detecting and identifying system faults and changes.

[1]  W.E. Larimore,et al.  Selecting the past and future for subspace identification of nonlinear systems with feedback and additive noise , 2006, 2006 American Control Conference.

[2]  Dietmar Bauer,et al.  Comparing the CCA Subspace Method to Pseudo Maximum Likelihood Methods in the case of No Exogenous Inputs , 2005 .

[3]  W.E. Larimore,et al.  Maximum likelihood subspace identification for linear, nonlinear, and closed-loop systems , 2005, Proceedings of the 2005, American Control Conference, 2005..

[4]  Wallace E. Larimore,et al.  Large Sample Efficiency for Adaptx Subspace System Identification with Unknown Feedback , 2004 .

[5]  Wallace E. Larimore,et al.  Inferring Multivariable Delay and Seasonal Structure for Subspace Modeling , 2003 .

[6]  W. E. Larimore,et al.  Automated multivariable system identification and industrial applications , 1999, Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).

[7]  Wallace E. Larimore,et al.  System Identification of Feedback and "Causality" Structure Using Canonical Variate Analysis , 1997 .

[8]  Wallace E. Larimore,et al.  Optimal Reduced Rank Modeling, Prediction, Monitoring and Control using Canonical Variate Analysis , 1997 .

[9]  Bart De Moor,et al.  A unifying theorem for three subspace system identification algorithms , 1995, Autom..

[10]  Manfred Deistler,et al.  Consistency and relative efficiency of subspace methods , 1994, Autom..

[11]  Michel Verhaegen,et al.  Identification of the deterministic part of MIMO state space models given in innovations form from input-output data , 1994, Autom..

[12]  John Baillieul,et al.  Identification and filtering of nonlinear systems using canonical variate analysis , 1993, Proceedings. The First IEEE Regional Conference on Aerospace Control Systems,.

[13]  Michèle Basseville,et al.  Detection of abrupt changes: theory and application , 1993 .

[14]  Clifford M. Hurvich,et al.  Bias of the corrected AIC criterion for underfitted regression and time series models , 1991 .

[15]  Wallace E. Larimore,et al.  Canonical variate analysis in identification, filtering, and adaptive control , 1990, 29th IEEE Conference on Decision and Control.

[16]  Wallace E Larimore,et al.  System Identification and Filtering of Nonlinear Controlled Markov Processes by Canonical Variate Analysis , 1989 .

[17]  W. E. Larimore,et al.  Order-recursive factorization of the pseudoinverse of a covariance matrix , 1988, Proceedings of the 27th IEEE Conference on Decision and Control.

[18]  W. Larimore.  System Identification, Reduced-Order Filtering and Modeling via Canonical Variate Analysis , 1983, 1983 American Control Conference.

[19]  Wallace E. Larimore,et al.  Predictive inference, sufficiency, entropy and an asymptotic likelihood principle , 1983 .

[20]  H. Hotelling.  Relations Between Two Sets of Variates , 1936 .

[21]  David R. Cox,et al.  PRINCIPLES OF STATISTICAL INFERENCE , 2017 .

[22]  Dale E. Seborg,et al.  Fault Detection Using Canonical Variate Analysis , 2004 .

[23]  Howell Tong,et al.  Non-Linear Time Series , 1990 .

[24]  H. Akaike.  Canonical Correlation Analysis of Time Series and the Use of an Information Criterion , 1976 .

[25]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .