摘要：AbstractAutomotive natural gas engines are designed to dramatically reduce the emission of polluting gases, particulate matter, and harmful substances associated to combustion of Diesel or gasoline fuels. Their performance depends on the accurate metering of the air/fuel ratio, which is pursued by controlling the gas injection timing and the pressure in the common rail volume. To this aim, an accurate control-oriented model is necessary to represent gas pressure dynamics in the main parts of the injection system. This work focuses on the gas transition from inner parts of the system to the common rail. The aim is to identify a linear non-integer-order model that describes the process in a more effective and compact way than ARX integer-order models of high order. Identification is performed in a non-parametric setting, trying to identify the optimal smooth fit in the frequency domain by minimization of an appropriate discrepancy criterion, between the actual measurements and the model outputs. In this way, the model structure will be obtained as part of the identification process and will not be fixed a priori.
关键词：KeywordsFractional-order systemsmodel identificationfrequency responseparticle swarm optimizationcompressed natural gas engines
