This paper newly proposes an on-line learning method for adaptation maps by using Gaussian filters. In this method, man-hour for calibration of adaptation maps can be decreased, and complicated structures of adaptation maps are learned without any prior knowledge. Moreover, by the effect of Gaussian filter, smoothed maps can be created even under noisy conditions or a few measured points. We also introduce improvements of the algorithm to cope with engine deterioration caused by aging. In this work, the proposed method is applied to minimum advance for best torque control on actual vehicles with just one driving date, and the accuracy of the learned map is verified through simulations and experiments.
各種検索
サポート
ヘルプ