Data-driven Soft Sensor Modeling for DCT Clutch Pressure Based on Decision Tree and Stepwise Regression

Abstract

Soft sensors have been widely used at fault diagnosis and active fault-tolerant control due to its price advantage. A data-driven model of soft sensor based on decision tree and stepwise regression algorithm (DT-SRA) is designed in this research to estimate the clutch pressure of dual-clutch transmission (DCT). The CART decision tree is designed to divide operating conditions into the process of shifting and operation of driving at a specific gear. Moreover, soft-sensor models of clutch pressure in different operating conditions are established by stepwise regression algorithm respectively. The proposed model is validated by multifarious datasets, that acquired by real-vehicle test. The results show that the Rsquared of predicting clutch pressure of the odd shaft and the even shaft can reach 97.5% and 95.2% respectively.