DOI QR코드

DOI QR Code

Detection Method of Vehicle Fuel-cut Driving with Deep-learning Technique

딥러닝 기법을 이용한 차량 연료차단 주행의 감지법

  • Ko, Kwang-Ho (Division of smart Automobile, Pyeongtaek University)
  • 고광호 (평택대학교 스마트자동차학과)
  • Received : 2019.10.10
  • Accepted : 2019.11.20
  • Published : 2019.11.28

Abstract

The Fuel-cut driving is started when the acceleration pedal released with transmission gear engaged. Fuel economy of the vehicle improves by active fuel-cut driving. A deep-learning technique is proposed to predict fuel-cut driving with vehicle speed, acceleration and road gradient data in the study. It's 3~10 of hidden layers and 10~20 of variables and is applied to the 9600 data obtained in the test driving of a vehicle in the road of 12km. Its accuracy is about 84.5% with 10 variables, 7 hidden layers and Relu as activation function. Its error is regarded from the fact that the change rate of input data is higher than the rate of fuel consumption data. Therefore the accuracy can be better by the normalizing process of input data. It's unnecessary to get the signal of vehicle injector or OBD, and a deep-learning technique applied to the data to be got easily, like GPS. It can contribute to eco-drive for the computing time small.

차량의 변속기어가 체결된 주행 상태에서 가속페달을 방치하는 경우 연료차단 주행이 시작된다. 적극적인 연료차단 주행을 활용하면 차량 연비가 개선된다. 본 연구에서는 차량의 속도, 가속도, 도로구배를 입력데이터로 사용하여 연료차단 주행 여부를 예측할 수 있는 딥러닝 기법을 제안하였다. 약 12km 정도의 도로주행을 통해 측정한 9600개의 데이터에 은닉층 3~10개, 매개변수 10~20개의 딥러닝 연산법을 적용하여 연료차단 주행여부를 예측하였다. 연산 결과, 렐루함수를 활성화함수로 적용하고 은닉층 7개, 매개변수 10개인 경우 정확도 84.5% 수준으로 예측할 수 있었다. 입력데이터인 속도, 가속도, 도로구배의 변화율이 연료소모율 데이터의 변화율에 비해 큰 것이 오차의 원인으로 판단된다. 따라서 입력데이터 정규화 과정을 통해 정확도를 높일 수 있을 것으로 예상된다. 본 연구의 특징은 차량의 연료분사 인젝터나 OBD 데이터를 사용하지 않고 GPS 등에서 쉽게 측정할 수 있는 데이터에 딥러닝을 적용한 방식이다. 또한 연산량이 적어 본 연구에서 제안한 방식으로 친환경 경제운전에 적용하기 용이할 것으로 기대된다.

Keywords

References

  1. M. Setiyo & S. Munahar. (2017). AFR and fuel cut-off modeling of LPG-fueled engine based on engine, transmission, and brake system using fuzzy logic controller (FLC). Journal of Mechatronics, Electrical Power and Vehicular Technology, 8(1), 50-59. DOI : 10.14203/j.mev.2017.v8.50-59.
  2. K. H. Ko & S. C. Choi. (2012). A study on the improvement of vehicle fuel economy by fuel-cut driving. JKAIS, 13(2), 498-503. UCI : G704-001653.2012.13.2.019
  3. S. C. Choi., K. H. Ko. & I. S. Jeung. (2013). Optimal fuel-cut driving method for better fuel economy. International Journal of Automotive Technology, 14(2), 183-187. DOI : 10.1007/s12239−013−0020−4
  4. K. H. Ko.. (2010). The Change Rate of Fuel Consumption for Different IRI of Paved Roads. Int. J. Highw. Eng., 12(1), 55-59. UCI : G704-001414.2010.12.1.009
  5. J. G. Yun. & J. Y. You. (2019). An Experimental Study on the Measurement of Fuel Consumption Using Fuel Flowmeter and CAN DATA. KSMT, 21(1), 63-68. DOI : 10.17958/ksmt.21.1.201902.63
  6. G. H. Lee., S. J. Lim. & S. J. O. (2017). Comparing the Effects of Visual and Visual-auditory Feedback on Eco-driving and Driving Workload. J. Korea Inst. Intell. Transp. Syst., 16(3), 120-131. DOI : 10.12815/kits.2017.16.3.120
  7. J. C. Park., J. H. Kim., W. T. O., J. H. Choi. & J. J. Park. (2017). Reliability Evaluation of EDR Data Using PC-Crash & Vbox. Transactions of KSAE, 25(3), 317-325. DOI : 10.7467/KSAE.2017.25.3.317
  8. S. H. Lee., I. S. Kim. & M. Robert. (2002). Model Reference Adaptive Control Using Non-Euclidean Gradient Descent. IJCAS, 4(4), 330-340. UCI : G704-000903.2002.4.4.010
  9. R. Ehsan. & E. Abbas. (2018). A Finite-time Adaptive Fuzzy Terminal Sliding Mode Control for Uncertain Nonlinear Systems. IJCAS, 16(4), 1938-1950. DOI : 10.1007/s12555-017-0552-x
  10. W. J. Lee., J. W. Lee. & S. R. Park. (2018). Instance Weighting Domain Adaptation Using Distance Kernel. IE&MS, 17(2), 334-340 DOI : 10.7232/iems.2018.17.2.334
  11. S. H. Rou. & J. B. Yun. (2017). The Effect of regularization and identity mapping on the performance of activation functions. JKAIS, 18(10), 75-80 DOI : 10.5762/KAIS.2017.18.10.75
  12. S. G. Inn. & J. H. Park. (2018). Inferring Method of Height and Weight from Smartphone Sensor Data Using Deep Learning and Clustering. KTCP, 24(9), 456-467 DOI : 10.5626/KTCP.2018.24.9.456
  13. J. K. Sung. & Y. K. Kim. (2017). Deep learning-based product image classification system and its usability evaluation for the O2O shopping mall platform. JIIBC, 17(3), 227-234 DOI : 10.7236/JIIBC.2017.17.3.227
  14. C. H. Peng, I. C. Yeh & L. C. Lien. (2009). Modeling strength of high-performance concrete using genetic operation trees with pruning techniques. CAC, 6(3), 203-223 UCI : G704-SER000008569.2009.6.3.005
  15. M. G. Ji., J. C. Chun. & N. G. Kim. (2018). An Improved Image Classification Using Batch Normalization and CNN. Journal of Internet and Services, 19(3), 35-42