http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
전자식 조향 장치에서 발생하는 실차 소음 예측을 위한 다중 센서 스펙트로그램 트랜스포머
고은지,남규환,김상욱,박경환,김성범 대한산업공학회 2022 대한산업공학회지 Vol.48 No.4
Noise level management has become an important task in the automotive industry because of the increased demand for low-noise automobiles. In particular, it is essential to reduce the automobile noises incurred by electric power steering(EPS) and an automotive steering system. Although existing methods measure the noise levels by mounting EPS on actual automobiles and detect the importance of frequency bands based on experts’ judgment, they were subjective, time consuming and expensive. Therefore, an efficient method is required to predict the automobile noise levels and detect influential frequency bands based on EPS acceleration data. We propose a multi-sensor spectrogram transformer(MuST) for predicting the EPS automobile noise levels. The proposed method allows us to predict automobile noise levels by reflecting the individual characteristics of multiple sensors in EPS and detecting influential frequency bands related to noise levels. The experimental results showed that the proposed MuST performed well and detected influential frequency bands similar to the experts. We believe that the framework presented in this study can efficiently identify the automobile noise levels and help experts design quiet automobiles in the future by providing information on influential frequency bands. In addition, our method can help to handle various tasks which use data collected through multiple sensors in the automotive industry.
차량 내부 소음 예측을 위한 멀티 모달 자기 지도 학습 네트워크
고은성,고은지,남규환,김상욱,박경환,김성범 한국경영과학회 2022 經營 科學 Vol.39 No.4
Predicting in-vehicle noise levels is an important issue in automobile industry. In most previous studies, various supervised learning methods that use both the input and output(labeled) variables are used to predict automobile noise levels. However, collecting labeled data for in-vehicle noise prediction is time consuming and expensive. In this study, we propose a multimodal self-supervised learning framework that can predict in-vehicle noise levels with only a small amount of labeled data, so that resources required to collect labeled data can be saved. In our framework, both original acceleration signals and spectrograms converted from the original data are used as the input to predict in-vehicle noise levels. In the first stage, we pretrain the features of the input data based on the relationship between the signal and spectrogram data using only unlabeled data, which is much easier to collect than labeled data. In the second stage, we use a small amount of data to construct the in-vehicle noise prediction model with the pretrained feature extractor. The effectiveness and applicability of the proposed framework are demonstrated using the actual acceleration signal data collected from various locations of electric power steering vehicle noise levels. The proposed framework outperforms the existing supervised learning method especially when a few labeled data are available.
다중 합성곱 모델을 이용한 R-EPS 차량 내부 소음 예측
고은성,정기원,조용원,고은지,안시후,남규환,김상욱,박경환,김성범 대한산업공학회 2022 대한산업공학회지 Vol.48 No.4
Because dealing with noises in automobile becomes more important, it is valuable to predict in-vehicle noise levels and use them for the product noise design. With the recent development of artificial intelligence, many studies have attempted to use deep learning models for various types of data generated in automobile industry. However, to the best of our knowledge, no studies have been conducted to predict in-vehicle noise levels based on deep learning models. In this study, we propose a deep learning framework that can predict in-vehicle noise levels and identify the causes of noises. Our framework is developed to recognize in-vehicle noise levels with automobile acceleration data from various locations of electric power steering devices. Our deep learning framework consists of several convolutional neural backbone networks to extract representation vectors for each acceleration axis. In addition, acceleration data are converted into a spectrogram through the short-term Fourier transformation technique, and high frequency bands in the spectrogram are removed to better represent the input data. We demonstrated that our proposed framework is suitable for predicting in-vehicle noises and identifying the major causes of noises. We expect that the explanation for prediction results will be helpful in the design low-noise vehicles.