Bayesian estimate of mass fraction of burned fuel in internal combustion engines using presssure measurements
Conference
Estumano, DC, Hamilton, FC, Colaço, MJ et al. (2014). Bayesian estimate of mass fraction of burned fuel in internal combustion engines using presssure measurements
. 997-1004.
Estumano, DC, Hamilton, FC, Colaço, MJ et al. (2014). Bayesian estimate of mass fraction of burned fuel in internal combustion engines using presssure measurements
. 997-1004.
The numerical simulation of combustion processes in internal combustion engines is a very difficult task. It involves the reacting turbulent flow of a gaseous mixture that compresses and burns in a short amount of time. Different models, with various levels of complexity, exist in the open literature and usually need calibration to work properly. Although being quite simple, a First Law analysis of this problem is widely used by the industry. Such formulation requires a model for the mass fraction of burned fuel, which is often based on theWiebe equation, and requires calibration using experimental data. The objective of this paper is to estimate the mass fraction of burned fuel using Bayesian particle filters. Particle filters, also called Sequential Monte Carlo (SMC) methods, fit into the domain of inverse modelling procedures, where measurements are incorporated into a computational model so as to formulate some feedback information on the uncertain model state variables and/or parameters, through accurate representations of their probability density functions. Based on a simple sampling importance distribution and resampling techniques, particle filters combine Monte Carlo samplings with sequential Bayesian filtering problems. In this particular application, measurements obtained from a pressure transducer located inside a combustion chamber are used to feed an observation model, while a First Law analysis is used as an evolution model to this Bayesian estimate. Very good results are obtained for the mass fraction of burned fuel, showing the great potential for this technique to be used as practical tool in the industry.
