A Systematic Approach to Support Drive Train Design Using Tools for Human Comfort Evaluation and Customer Classification

Abstract

This article presents a systematic procedure and Artificial Neural Networks (ANN) based tool for comfort objectification and customer classification, to support drive train developer during the product development process. In this case, the term “comfort objectification” can be clarified as reproduction of subjectively sensed convenience of a passenger through objectively measurable values. Many factors, such as noise, vibration, physical or psychological condition of a passenger generally influence the ride comfort. The main purpose of this project is to develop the drive train and his assemblies which can sustain customers’ demand of vibration comfort. The presented methods enable the identification and the evaluation of vehicle dynamic properties from the passengers’ point of view during start-up, shifting, steering as well as other procedures in the early stage of the product development process. For instance, this tool is developed for the evaluation of ride comfort during a start-up of a front-drive, intermediate-class car. To estimate the subjective sensation, the new driver modeling tool based on ANN is developed from the way individual drivers make their assessment. This paper presents a user-friendly interface which allows both experts and users who are still short of experience in the ANN field, to create different network structures depending on the task. By means of this tool, the modeling process can be effectively simplified and shortened. As a result, the objective values captured during each drive test are efficiently correlated with the subjective sensation. Consequently, the high performance of comfort prediction can be achieved. By using self organizing map as a tool for driver classification, the different types of drivers can be considered due to their comfort expectation and style of driving. The comfort prediction concerning each driver group can then be carried out. According to the approach of virtual drive train development, in this study, the elaborated multi-body simulation models are primarily used to generate several virtual start-up processes. This enables the determination of NVH properties of the future product and allows the developer to investigate the effect of vibration like judder and jerking on the degree of ride comfort. By applying objective data from the simulation, the comfort assessment using the presented tools can be executed. In the long run, costly drive tests and cost-intensive prototypes can be partially replaced.