Dynamic engine modeling through linear programming Support Vector Regression

In this paper, we develop a dynamic model for an internal combustion engine using Support Vector Regression (SVR). In particular, a linear programming SVR (LP-SVR) approach is investigated. The computational advantages and generalization capability of the LP-SVR dynamic engine model are illustrated through a case study, where a model is developed for an L4 gasoline engine. Simulation results are reported to demonstrate the effectiveness of proposed approach and to illustrate the trade-offs among different modeling attributes.