RESUMEN

This study researched the prediction of the BSR noise evaluation quantitative index, Loudness N10, for sound sources with noise using statistics and machine learning. A total of 1170 data points was obtained from 130 automotive seats measured at 9-point positions, with Gaussian noise integrated to construct synthetic sound data. Ten physical quantities related to sound quality and sound pressure were used and defined as dB and fluctuation strength, considering statistical characteristics and Loudness N10. BSR quantitative index prediction was performed using regression analysis with K-fold cross-validation, DNN in hold-out, and DNN in K-fold cross-validation. The DNN in the K-fold cross-validation model demonstrated relatively superior prediction accuracy, especially when the data quantity was relatively small. The results demonstrate that applying machine learning to BSR prediction allows for the prediction of quantitative indicators without complex formulas and that specific physical quantities can be easily estimated even with noise.

RESUMEN

This study investigated the squeal mechanism induced by friction in a lead screw system. The dynamic instability in the friction noise model of the lead screw was derived through a complex eigenvalue analysis via a finite element model. A two degree of freedom model was described to analyze the closed solutions generated in the lead screw, and the friction noise sensitivity was examined. The analysis showed that the main source of friction noise in the lead screw was the bending mode pair, and friction-induced instability occurred when the ratio of the stiffness of the bending pair modes was 0.9-1. We also built an architecture to predict multiple outputs from a single model using deep neural networks and demonstrated that friction-induced instability can be predicted by deep neural networks. In particular, instability with nonlinearity was predicted very accurately by deep neural networks with a maximum absolute difference of about 0.035.

Asunto(s)

RESUMEN

In this paper, an optimal design model was developed to reduce noise and secure the torque performance of a brushless direct-current motor used in the seat of an autonomous vehicle. An acoustic model using finite elements was developed and verified through the noise test of the brushless direct-current motor. In order to reduce noise in the brushless direct-current motor and obtain a reliable optimization geometry of noiseless seat motion, parametric analysis was performed through the design of experiments and Monte Carlo statistical analysis. The slot depth, stator tooth width, slot opening, radial depth, and undercut angle of the brushless direct-current motor were selected as design parameters for design parameter analysis. Then, a non-linear prediction model was used to determine the optimal slot depth and stator tooth width to maintain the drive torque and minimize the sound pressure level at 23.26 dB or lower. The Monte Carlo statistical method was used to minimize the deviation of the sound pressure level caused by the production deviation of the design parameters. The result is that the SPL was 23.00-23.50 dB with a confidence level of approximately 99.76% when the level of production quality control was set at 3σ.

RESUMEN

The chaotic squeak and rattle (S&R) vibrations in mechanical systems were classified by deep learning. The rattle, single-mode, and multi-mode squeak models were constructed to generate chaotic S&R signals. The repetition of nonlinear signals generated by them was visualized using an unthresholded recurrence plot and learned using a convolutional neural network (CNN). The results showed that even if the signal of the S&R model is chaos, it could be classified. The accuracy of the classification was verified by calculating the Lyapunov exponent of the vibration signal. The numerical experiment confirmed that the CNN classification using nonlinear vibration images as the proposed procedure has more than 90% accuracy. The chaotic status and each model can be classified into six classes.

Asunto(s)