http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
생성적 적대 신경망 기반의 CCU 자동화 평가 강화 방법
이재영(Jaeyoung Lee),최두일(Dooil Choi),이성구(Sunggu Yi),이재환(Jaehwan Lee) 한국자동차공학회 2020 한국자동차공학회 부문종합 학술대회 Vol.2020 No.7
The number of controllers fitted continues to increase as the functional requirements of the vehicle become more advanced. Controller area network (CAN), CAN with flexible data rate and Ethernet are used to exchange information between controllers. Central communication unit (CCU) was developed to manage the entire network and the information transmission between hybrid networks. CCU is capable of connecting the vehicle"s internal and external networks using modems as well as the internal network. Therefore, vehicle to everything (V2X), over the air (OTA) and vehicle remote control functions can also be fused. The difficulty of system evaluation also increases because the functions of CCU become more complex. CCU is connected to all communication networks in the vehicle and receives hundreds of inputs via communication. If only certain signal values are monitored, it is easy to know whether they are normal or not. However, it is difficult to intuitively check whether the output values are normal when the entire input changes at the same time. In this paper, we propose a CCU automation assessment reinforcement method based on generative adversarial neural network (GAN) to overcome the limitations of rulebased system evaluations. The proposed system uses time series GAN to generate input signals similar to the actual vehicle. In addition, abnormality detection method is used to determine probability of abnormalities in CCU. To verify the proposed method, by using signals acquired from actual vehicles, an experiment is conducted to determine if there are any new problems by regenerating the input signal of CCU. It improves CCU system evaluation reliability since it increases test coverage a lot.