Intelligent Gear Decision Method for Automatic Vehicles Based on Data Mining Under Uphill Conditions

Abstract

Existing gear decision methods are affected by the limiting factors of deviation of numerical models as well as the subjectivity of the designer, which results in poor adaptability to the driving intentions and driving environments. Excellent drivers are able to choose the appropriate gear to satisfy their driving intentions and respond to changes in driving environments when driving manual transmission vehicles. Therefore, an intelligent gear decision method and design methodology of vehicle automatic transmission system that adapts to different driving intentions and complex driving environments are constructed from the massive driving data of excellent drivers by using data mining method. Excellent drivers are employed to drive vehicles with manual transmission to obtain a large amount of driving data. Subsequently, data preprocessing, data cleaning, and outlier removal are performed on the collected driving data to extract the shift boundary points of each gear aiming at constructing the shift rule surface of each gear. Using the uphill condition as an example, a data-mining-based shift control strategy is established, and the comparison results verify that the proposed gear decision design method can mine the shift strategy of excellent drivers from massive driving data, and the constructed strategy can better adapt to the driver’s intention while obtaining better fuel economy and avoiding unreasonable gearshifts compared with the automatic transmission’s default strategy.