Parameter estimation in non-linear models of pressure dynamics in CNG injection systems

Abstract

Common rail injection systems in innovative gas engines require an accurate control of the rail pressure dynamics. This task is, in fact, combined with a precise setting of the opening/closing time intervals of electro-injectors to achieve a satisfactory metering of the ratio between gas and air, then a good performance that reduces pollution and consumption. To achieve a trade-off between model accuracy and control simplicity, an important issue is to optimize the model of the pressure dynamics in the main accumulation volumes. This paper reports preliminary results in estimating the optimal parameters of a non-linear model by two different techniques. First delays in the main variables are estimated by a particle swarm optimization. Then, time-variability of model parameters is investigated.