# Generating Component Designs for an Improved NVH Performance by Using an Artificial Neural Network as an Optimization Metamodel

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Approach

#### 2.1. Use Case Scenario

#### 2.2. Workflow

#### 2.3. Artificial Neural Network

#### 2.4. Creation of the Target FRF

#### 2.5. Obtaining the New Parameter Set

#### 2.6. Process Verification

## 3. Application Examples

#### 3.1. Target FRF with Amplitude Reduction

#### 3.2. Maximum-Curve with Multiple Amplitude Optimization Targets

#### 3.3. Change of Eigenfrequency

#### 3.4. Combined Amplitude and Frequency Changes

#### 3.5. Comparison to Earlier Investigations

## 4. Conclusions and Outlook

## Author Contributions

## Funding

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## Abbreviations

ANN | Artificial Neural Network |

FE | Finite Element |

FEM | Finite Element Method |

FRF | Frequency Response Function |

HL | Hidden Layer |

MSE | Mean Squared Error |

NVH | Noise, Vibration, and Harshness |

ReLu | Rectified Linear Unit |

SFLH | Space Filling Latin Hypercube |

## References

- Jung, F. 5th International ATZ Conference on Vehicle Acoustics. ATZ Worldw.
**2019**, 121, 76–77. [Google Scholar] [CrossRef] - Mantovani, M. Rolling noise does not evoke an emotional response. ATZ Worldw.
**2018**, 120, 18–21. [Google Scholar] [CrossRef] - Rambacher, C.; Ehrt, T.; Sell, H. Vibration optimisation of entire axles. ATZ Worldw.
**2017**, 119, 50–55. [Google Scholar] [CrossRef] - Uhlar, S. Hybrid Simulation of Structure-borne Road Noise. ATZ Worldw.
**2021**, 123, 50–55. [Google Scholar] [CrossRef] - Hua, X.; Thomas, A.; Shultis, K. Recent progress in battery electric vehicle noise, vibration, and harshness. Sci. Prog.
**2021**, 104, 368504211005224. [Google Scholar] [CrossRef] [PubMed] - Schlecht, A. Minimierung der Schwingungsempfindlichkeit von Kraftfahrzeugvorderachsen. Ph.D. Thesis, Technische Universität München, München, Germany, 2012. [Google Scholar]
- Schlott, S. Endless conflicts. ATZ Worldw.
**2018**, 120, 6–7. [Google Scholar] [CrossRef] - Ichikawa, K. Research on Mechanism of Change in Suspension Transfer Force in Relation to Low-Frequency Road Noise. SAE Int. J. Passeng. Cars Mech. Syst.
**2015**, 8, 22–30. [Google Scholar] [CrossRef] - Arana, C.; Evangelou, S.A.; Dini, D. Series Active Variable Geometry Suspension for Road Vehicles. IEEE/ASME Trans. Mech.
**2015**, 20, 361–372. [Google Scholar] [CrossRef] [Green Version] - Ersoy, M.; Gies, S. (Eds.) Fahrwerkhandbuch: Grundlagen, Fahrdynamik, Komponenten, Systeme, Mechatronik, Perspektiven, 5th ed.; ATZ/MTZ-Fachbuch, Springer Vieweg: Wiesbaden, Germany, 2017. [Google Scholar] [CrossRef]
- Khan, S.; Horoub, M.M.; Shafiq, S.; Ali, S.; Bhatti, U.N. Optimization of Vehicle Suspension System Using Genetic Algorithm. In Proceedings of the 2019 IEEE 10th International Conference on Mechanical and Aerospace Engineering (ICMAE), Brussels, Belgium, 22–26 July 2019; pp. 203–207. [Google Scholar] [CrossRef]
- Kosaka, F.; Mizuno, H.; Inoue, T.; Takagi, K. Road noise sensitivity analysis with respect to suspension geometry. In Proceedings of the INTER-NOISE and NOISE-CON Congress and Conference Proceedings, Melbourne, Australia, 16–19 November 2014; Volume 249, pp. 2127–2136. [Google Scholar]
- Harzheim, L. Strukturoptimierung: Grundlagen und Anwendungen; 3. überarbeitete und Erweiterte Auflage ed.; Deutsch, H., Ed.; Verlag Europa Lehrmittel—Nourney Vollmer GmbH & Co. KG: Haan-Gruiten, Germany, 2019. [Google Scholar]
- von Wysocki, T.; Chahkar, J.; Gauterin, F. Small Changes in Vehicle Suspension Layouts Could Reduce Interior Road Noise. Vehicles
**2020**, 2, 18–34. [Google Scholar] [CrossRef] [Green Version] - Reff, B.; Palandri, J.; Rambacher, C.; Sell, H. Chassis system NVH: Advanced analysis methods—Full vehicle correlation. In Automotive Acoustics Conference; Springer: Berlin, Germany, 2015. [Google Scholar]
- TrelleborgVibracoustic. Schwingungstechnik im Automobil: Grundlagen, Werkstoffe, Konstruktion, Berechnung und Anwendungen, 1st ed.; Vogel Business Media: Würzburg, Germany, 2015. [Google Scholar]
- Marburg, S. Efficient optimization of a noise transfer function by modification of a shell structure geometry—Part I: Theory. Struct. Multidiscip. Optim.
**2002**, 24, 51–59. [Google Scholar] [CrossRef] - Helm, D.; Huf, A.; Zimmer, H.; Kondziella, R. Anforderungen in der frühen Phase der Gesamtfahrzeugauslegung. In VDI-Berichte 2169; VDI-Verlag: Düsseldorf, Germany, 2012; pp. 21–40. [Google Scholar]
- Vemireddy, K.; Dittmar, T.; Eckstein, L.; Hesse, L.; Rettweiler, P. Development of a driving dynamics-oriented suspension design during the early concept phase. In 6th International Munich Chassis Symposium 2015; Pfeffer, P.E., Ed.; Springer Fachmedien Wiesbaden: Wiesbaden, Germany, 2015; pp. 233–255. [Google Scholar] [CrossRef]
- Genuit, K. (Ed.) Sound-Engineering im Automobilbereich: Methoden zur Messung und Auswertung von Geräuschen und Schwingungen; Springer: Berlin/Heidelberg, Germany, 2010. [Google Scholar] [CrossRef]
- Arslan, M.A.; Hajela, P. Use of counterpropagation neural networks to enhance the concurrent subspace optimization strategy. Eng. Optim.
**2001**, 33, 327–349. [Google Scholar] [CrossRef] - Most, T.; Will, J. Sensitivity analysis using the Metamodel of Optimal Prognosis. In Proceedings of the Weimar Optimization and Stochastic Days 8, Weimar, Germany, 24–25 November 2011; pp. 24–40. [Google Scholar]
- Niersmann, A. Modellbasierte Fahrwerkauslegung und -Optimierung. Ph.D. Thesis, Technische Universität Braunschweig, Braunschweig, Germany, 2012. [Google Scholar]
- Ranjbar, M.; Marburg, S. Fast Vibroacoustic Optimization of Mechanical Structures Using Artificial Neural Networks. Int. J. Mech. Eng. Appl.
**2013**, 1, 64. [Google Scholar] [CrossRef] - Siebertz, K.; van Bebber, D.; Hochkirchen, T. Statistische Versuchsplanung; Springer: Berlin/Heidelberg, Germany, 2017. [Google Scholar] [CrossRef]
- Büttner, J.; Schwarz, S.; Schumacher, A.; Bäck, T. Global Sensitivity Matrix for Vehicle Development. ATZ Worldw.
**2021**, 123, 26–31. [Google Scholar] [CrossRef] - von Wysocki, T.; Leupolz, M.; Gauterin, F. Metamodels Resulting from Two Different Geometry Morphing Approaches Are Suitable to Direct the Modification of Structure-Born Noise Transfer in the Digital Design Phase. Appl. Syst. Innov.
**2020**, 3, 47. [Google Scholar] [CrossRef] - Nagaya, K.; Li, L. Control of sound noise radiated from a plate using dynamic absorbers under the optimization by neural network. J. Sound Vib.
**1997**, 208, 289–298. [Google Scholar] [CrossRef] - Shi, Q.; Hagiwara, I.; Azetsu, S.; Ichkawa, T. Holographic neural network approximations for acoustics optimization. JSAE Rev.
**1998**, 19, 361–363. [Google Scholar] [CrossRef] - Tsokaktsidis, D.E.; von Wysocki, T.; Gauterin, F.; Marburg, S. Artificial Neural Network predicts noise transfer as a function of excitation and geometry. In Proceedings of the 23rd International Congress on Acoustics, Aachen, Germany, 9–13 September 2019; pp. 4378–4382. [Google Scholar] [CrossRef]
- Irvine, T. An Introduction to Frequency Response Functions; College of Engineering and Computer Science, California State University: Northridge, CA, USA, 2000. [Google Scholar]
- Ansys. Ansys optiSLang. 2021. Available online: https://www.ansys.com/products/platform/ansys-optislang (accessed on 2 June 2021).
- Cybenko, G. Approximation by superpositions of a sigmoidal function. Math. Control. Signals Syst.
**1989**, 2, 303–314. [Google Scholar] [CrossRef] - de Myttenaere, A.; Golden, B.; Grand, B.L.; Rossi, F. Mean Absolute Percentage Error for regression models. Neurocomputing
**2016**, 192, 38–48. [Google Scholar] [CrossRef] [Green Version] - LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature
**2015**, 521, 436–444. [Google Scholar] [CrossRef] [PubMed] - Glorot, X.; Bengio, Y. Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the JMLR Workshop and Conference Proceedings, Chia Laguna Resort, Sardinia, Italy, 13–15 May 2010; Volume 9, pp. 249–256. [Google Scholar]
- Glorot, X.; Bordes, A.; Bengio, Y. Deep sparse rectifier neural networks. In Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, Fort Lauderdale, FL, USA, 11–13 April 2011; pp. 315–323. [Google Scholar]
- Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning; MIT Press: Cambridge, MA, USA, 2016. [Google Scholar]
- Hadka, D. Platypus. 2020. Available online: https://github.com/Project-Platypus/Platypus (accessed on 2 June 2021).
- Free Software Foundation. GNU General Public License v3.0. Available online: https://www.gnu.org/licenses/gpl-3.0.html (accessed on 2 June 2021).
- Civita, M.L.; Sestieri, A. Optimization of an Engine Mounting System for Vibro-Acoustic Comfort Improvement. In Proceedings of the SPIE The International Society for Optical Engineering, 1998; Volume 3727/PT2, pp. 1998–2004. Available online: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.458.6086&rep=rep1&type=pdf (accessed on 2 June 2021).

**Figure 1.**Knuckle of a vehicle suspension. The flat reddish surface in the center represents the wheel hub attachment surface. Gray spheres inside gray surfaces represent variable kinematic hard points. FRFs describe the noise transfer between each blue arrow and each red arrow. Each arrow represents one of the three directions ${x}_{1}$, ${x}_{2}$ or ${x}_{3}$.

**Figure 2.**Flow chart representing the presented approach. The simulated designs created via DoE serve as training data for multiple ANNs. The optimizer identifies fitting parameter sets that fulfill the target curves. The identified design is then validated.

**Figure 3.**User input for the creation of a target curve. A graphical application presents the original FRF (black) and the skyline of all FRFs of the DoE data set (gray). The design engineer draws (pen symbol) a desired target curve into the plot (blue).

**Figure 5.**Optimization using a matching-curve for a desired amplitude reduction around 1700 $\mathrm{Hz}$.

**Figure 6.**Optimization using a drawn maximum-curve describing a multi-objective optimization. Desired amplitude reduction at multiple frequency ranges while allowing small eigenfrequency shifts.

**Figure 7.**Optimization using a matching-curve to achieve an eigenfrequency raise at 1700 $\mathrm{Hz}$.

**Figure 8.**Optimization using a matching-curve to achieve an eigenfrequency reduction at 1900 $\mathrm{Hz}$.

**Figure 9.**Optimization using a matching-curve to achieve an eigenfrequency raise and an amplitude reduction at 1350 $\mathrm{Hz}$.

**Figure 10.**Optimization using a matching-curve to achieve an eigenfrequency reduction and an amplitude reduction at 1350 $\mathrm{Hz}$.

**Figure 11.**Optimization using a maximum-curve to achieve an eigenfrequency and amplitude reduction at multiple frequency ranges.

Layer | Size | Activation |
---|---|---|

In Layer ${d}_{\mathrm{IN}}$ | 15 | – |

Layer 1 ${d}_{\mathrm{HL}1}$ | 2^{10} | ReLU |

Layer 2 ${d}_{\mathrm{HL}2}$ | 2^{10} | ReLU |

Layer 3 ${d}_{\mathrm{HL}3}$ | 2^{10} | ReLU |

Out Layer ${d}_{\mathrm{OUT}}$ | 1000 | Identity |

Parameter | Value |
---|---|

Batch size | 10 |

Epochs | 200 |

Optimizer | Adam |

Loss | MSE |

Dropout ${d}_{\mathrm{HL}1}$, ${d}_{\mathrm{HL}2}$ | 20% |

Dropout ${d}_{\mathrm{HL}3}$ | 40% |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

von Wysocki, T.; Rieger, F.; Tsokaktsidis, D.E.; Gauterin, F.
Generating Component Designs for an Improved NVH Performance by Using an Artificial Neural Network as an Optimization Metamodel. *Designs* **2021**, *5*, 36.
https://doi.org/10.3390/designs5020036

**AMA Style**

von Wysocki T, Rieger F, Tsokaktsidis DE, Gauterin F.
Generating Component Designs for an Improved NVH Performance by Using an Artificial Neural Network as an Optimization Metamodel. *Designs*. 2021; 5(2):36.
https://doi.org/10.3390/designs5020036

**Chicago/Turabian Style**

von Wysocki, Timo, Frank Rieger, Dimitrios Ernst Tsokaktsidis, and Frank Gauterin.
2021. "Generating Component Designs for an Improved NVH Performance by Using an Artificial Neural Network as an Optimization Metamodel" *Designs* 5, no. 2: 36.
https://doi.org/10.3390/designs5020036