Stochastic transfer function in Bayesian inference for combustion indicator estimation

This paper deals with the use of Bayesian inference to regularize an inverse problem with a non linear transfer function. Bayesian inference is an useful tool to combine a statistical approach coming from a set of sensors and prior physical models of the system. Bayesian inference is applied to the estimation of combustion indicators. This is an inverse problem where only an indirect measurement from engine block vibrations is available and combustion models are necessary to extract relevant combustion parameters. Moreover the engine block is not a linear system and Bayesian inference can take into account this non linearity.