Automated Model Initialization Using Test Data

摘要

Building a vehicle model with sufficient accuracy for fuel economy analysis is a time-consuming process, even with the modern-day simulation tools. Obtaining the right kind of data for modeling a vehicle can itself be challenging, given that while OEMs advertise the power and torque capability of their engines, the efficiency data for the components or the control algorithms are not usually made available for independent verification. The U.S. Department of Energy (DOE) funds the testing of vehicles at Argonne National Laboratory, and the test data are publicly available. Argonne is also the premier DOE laboratory for the modeling and simulation of vehicles. By combining the resources and expertise with available data, a process has been created to automatically develop a model for any conventional vehicle that is tested at Argonne. This paper explains the process of analyzing the publicly available test data and computing the parameters of various components from the analysis. It includes the development of engine fuel maps, gear shift maps, and other vehicle specifications necessary to build a model. The paper also includes a case study in which this process is applied on test data from a conventional vehicle, and compares the results from the actual test to the outputs from the automatically generated vehicle model. This analysis shows the advantages and limitations of the process.