A modeling approach of human sensation based on ANN (artificial neural networks) was presented and validation results were presented exemplarily. It has been shown that ANN modelling is capable of adapting to different boundary conditions as the may occur by applying the IPEK-XiL (X-in-the-loop) approach on different validation levels or test benches. Thus, it already became possible to perform vehicle comfort and driveability assessment on Vehicle-in-the-Loop roller test benches without significant loss of accuracy. This is heavily influenced by the ability of adequate Rest-Vehicle-Models according to XiL. Since the applied method characterizes and objectivizes the values adequately for vehicle start-up it is possible to reproducibly evaluate the customer sensation and acceptance. In order to extend the field of usage this approach is about to be transferred to gear shifting comfort optimization aspects. In addition, the correlation ability of psycho-acoustic-related aspects respectively phenomena with human sensation was presented in detail. The results were performed based on long-term experience in modeling, testing and simulation for automotive validation at IPEK. ANN modeling, training, optimization and application was consequently carried out by using the GUI-based IPEK Driver Modeling Tool. Due to the elementary simplification of modeling net structures with highest performance possible have been chosen as well as unintended mistakes have been reduced to a minimum since effort for programming was minimized. The results underline the potential for identification of characteristic parameters including psycho-acoustics that have significant influence on human sensation and consequential rating with respect to the diversity of today´s and prospective drive systems. As well, the strong link of XiL enabling ANN integration on all validation levels finally provides approaches and methods for vehicle comfort assessment already in early development phases. Therefore another contribution to human sensation modeling by means of ANN was investigated, validated and presented.


    Zugriff

    Zugriff über TIB

    Verfügbarkeit in meiner Bibliothek prüfen


    Exportieren, teilen und zitieren



    Titel :

    Comfort objectivation for NVH and acoustics by means of artificial neural networks


    Beteiligte:
    Albers, A. (Autor:in) / Behrendt, M. (Autor:in) / Ott, S. (Autor:in)


    Erscheinungsdatum :

    2011


    Format / Umfang :

    17 Seiten, 12 Bilder, 11 Quellen


    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Datenträger


    Sprache :

    Englisch




    Comfort objectivation for NVH and acoustics by means of artificial neural networks

    Albers,A. / Behrendt,M. / Ott,S. et al. | Kraftfahrwesen | 2011


    Prediction of Subjective Comfort Ratings by Means of Artificial Neural Networks

    Albers, A. / Albrecht, M. / VDI Gesellschaft Fahrzeug- und Verkehrstechnik | British Library Conference Proceedings | 2002


    Thermal sensations and artificial neural networks: application to automotive car comfort

    Hernandez-Gress, N. / Esteve, D. / Piniec, S. et al. | British Library Conference Proceedings | 1997


    Thermal sensations and artificial neural networks: application to automotive car comfort

    Hermandez-Gress,N.M / Piniec,S. / Esteve,D. et al. | Kraftfahrwesen | 1997


    Using Artificial Neural Networks as a Tool to Estimate Vehicle Ride Comfort

    Stammen, K. / Meywerk, M. / VDI-Gesellschaft Fordertechnik, Materialfluss und Logistik | British Library Conference Proceedings | 2007