Search Results

This paper considers using linear quadratic regulation (LQR) for multi-input control of the Automatic Transmission (AT) upshift inertia phase. The considered control inputs include the transmission input/engine torque, oncoming clutch torque, and traditionally not used off-going clutch torque. Use of the off-going clutch has been motivated by discussed Control Trajectory Optimization (CTO) results demonstrating that employing the off-going clutch during the inertia phase along with the main, oncoming clutch can improve the upshift control performance in terms of the shift duration and/or comfort by trading off the transmission efficiency and control simplicity to some extent. The proposed LQR approach provides setting an optimal trade-off between the conflicting criteria related to driving comfort and clutches thermal energy loss.

Automatic transmission (AT) upshift control performance in terms of shift duration and comfort can be improved during the inertia phase by coordinating the off-going clutch together with oncoming clutch and engine torque. The performance improvement is highest in low gear shifts (i.e., for high ratio steps), which are typically performed with open torque converter. In this paper, a discrete-time, linear quadratic regulation (LQR) is applied during the upshift inertia phase, as it provides an optimal multi-input/multi-output control action with respect to the prescribed cost function. The LQR law is based on a reduced-order drivetrain model, which is applicable to actual transmissions characterized by a limited number of available state measurements. The reduced-order model includes the linearized torque converter model. The shift duration is ensured by precise tracking of a linear-like oncoming clutch slip speed reference profile.

The advancement of Machine-learning (ML) methods enables data-driven creation of Reduced Order Models (ROMs) for automotive components and systems. For example, Gaussian Process Regression (GPR) has emerged as a powerful tool in recent years for building a static ROM as an alternative to a conventional parametric model or a multi-dimensional look-up table. GPR provides a mathematical framework for probabilistically representing complex non-linear behavior. Today, GPR is available in various programing tools and commercial CAE packages. However, the application of GPR is system dependent and often requires careful design considerations such as selection of input features and specification of kernel functions. Hence there is a need for GPR design optimization driven by application requirements. For example, a moving window size for training must be tuned to balance performance and computational efficiency for tracking changing system behavior.