Tractable model development and system identification for longitudinal vehicle dynamics

A. Ganguli, R. Rajamani

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


Longitudinal vehicle models that include engine and powertrain dynamics tend to be either very detailed or empirical. Detailed cycle-by-cycle engine models, for instance, are of high order and require knowledge of a large number of parameters. Empirical models, on the other hand, are tractable but have to be obtained by extensive dynamometer testing. In this paper a spark ignition engine model is presented that does not require extensive dynamometer testing and at the same time is tractable enough to be useful for adaptive cruise control, speed control, diagnostics and other longitudinal control applications. The model is obtained by modifying existing dynamic models from the literature through a set of reasonable approximations so as to obtain a model that is amenable to parameter identification. The parameter identification is based on a least-squares error algorithm and utilizes sensor measurements obtained directly from on-the-vehicle road tests. The model and the parameter identification technique are experimentally evaluated using experimental data from a three-car platoon. The cars in the three-vehicle platoon are identical. The vehicle parameters are identified using the data from the lead car while data from cars 2 and 3 serve as experimental validation for the identified parameters. Results show that the model developed in the paper provides good estimates of the state variables except at discontinuous speed variations that occur during gear changes.

Original languageEnglish (US)
Pages (from-to)1077-1084
Number of pages8
JournalProceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
Issue number10
StatePublished - Oct 2004


  • Longitudinal vehicle dynamics
  • Spark ignition engine
  • Tractable model

Fingerprint Dive into the research topics of 'Tractable model development and system identification for longitudinal vehicle dynamics'. Together they form a unique fingerprint.

Cite this