A Modified Monte-Carlo Approach to Simulation-Based Vehicle Parameter Design with Multiple Performance Objectives and Multiple Scenarios

Shorter development times in the automotive industry are leading to the increased use of computer simulation in the vehicle design cycle to pre-optimize vehicle concepts. The focus of the work presented in this study is vehicle dynamic performance in different driving maneuvers. More specifically this paper presents a methodology for simulation-based parameter design of vehicles for excellent performance in multiple maneuvers. The model used in the study consists of eight degrees-of-freedom and has been validated previously. The vehicle data used is for a commercially available vehicle. A number of different driving scenarios (maneuvers) based on ISO standards for transient dynamic behavior are implemented and performance indices are calculated for each individual maneuver considered. Vehicle performance is assessed based on the performance indices.

Increasing the number of maneuvers increases the dimensionality of the design problem due to the performance indices associated with each maneuver. In order to reduce dimensionality, statistical methods are used to identify correlations between the performance indices and thereby reduce the number of performance indices. The methodology consists of changing various vehicle design parameters in a systematic way until an improvement in performance indices is achieved. A Latin Hypercube-based Monte-Carlo optimization method is then used to identify the ‘best’ points in the considered design space.

The results show that significant improvement in dynamic performance indices (up to 19% in some cases) can be obtained by the method utilized in this research. Further studies in the use of evolutionary algorithms for vehicle design are being initiated.