Title: Reducing Amesim models in the form af a recursive neural network
Renault and Siemens have developed a process and tool chain to reduce AMESim models in the form of a recursive neural network. Obtaining non-linear models of AMESim models that simulate very quickly and from which one can very easily extract linear equivalent model is a real challenge. This allows to open incredible gate for control system side, since from a model 1D AMESim, we are able to obtain systematically the embedded models for ECU. We will illustrate this tools chain through examples on internal combustion engines for the synthesis of virtual sensors and advanced control type MPC.