Simple Engine Exhaust Temperature Modeling and System Identification Based on Markov Chain Monte Carlo

2014 ◽  
Vol 598 ◽  
pp. 224-228
Author(s):  
Zheng Mao Ye ◽  
Habib Mohamadian

Even though actual composition of engine exhaust gases varies across diverse types of engines, such as gasoline, diesel, gas turbine and natural gas engines, engine exhaust temperature is always a major factor with strong impact on emission levels and catalytic converting efficiency. For spark ignition engines, exhaust temperature depends on various engine parameters, such as engine speed, engine load, A/F ratio, intake air temperature, coolant temperature and spark timing, etc. Due to complexity, it is impossible to share a unique analytical model of engine exhaust temperature. Instead, it is mostly modeled as a complicated nonlinear system. The model complexity increases significantly however accuracy cannot be guaranteed. On the other hand, a simple linear model with accurate system identification could serve as a versatile alternative to represent the engine exhaust temperature, while engine parameters are subject to model identification to be adaptable across different types of engines. Combination of linear functions in terms of dominant engine parameters of engine speed and engine load is used for exhaust temperature modeling. To identify optimal parameters, Markov Chain Monte Carlo (MCMC) is applied. The discrete-time Markov chain is introduced where the stationary probability replaces posterior density in Monte Carlo integration for numerical integration. Compared with the high order nonlinear approaches, low computation cost is involved in the simplified model. Good agreement between the model prediction data and testing results is observed. The approach could be easily extended to other types of engines.

Author(s):  
P. L. Green ◽  
K. Worden

In this paper, the authors outline the general principles behind an approach to Bayesian system identification and highlight the benefits of adopting a Bayesian framework when attempting to identify models of nonlinear dynamical systems in the presence of uncertainty. It is then described how, through a summary of some key algorithms, many of the potential difficulties associated with a Bayesian approach can be overcome through the use of Markov chain Monte Carlo (MCMC) methods. The paper concludes with a case study, where an MCMC algorithm is used to facilitate the Bayesian system identification of a nonlinear dynamical system from experimentally observed acceleration time histories.


Sign in / Sign up

Export Citation Format

Share Document