Intelligent robust control of clutch-to-clutch shifts in vehicle transmission systems

In a recently synthesized class of transmissions, the freewheelers are not present and clutch-to-clutch shift, a non-trivial control problem, is required. In controlling vehicle transmissions with clutch-to-clutch shifts, there are different types of uncertainties, actuator delay and inconsistency, actuator static errors, actuator dynamics, plant modeling errors, unknown external disturbance, and plant structure variations, that the controller has to overcome. An intelligent robust control system is developed to compensate for these uncertainties and achieve consistent and satisfactory clutch-to-clutch shifts.; The major achievements of this thesis can be summarized as follows: (1) A unique powertrain model for clutch-to-clutch shifts is developed. A checking logic is designed to examine the lock-up conditions of the clutches, and different equations are used for different modes of operation. (2) A Hybrid control algorithm consists of a Rule-Based Module and a Sliding Control Module is synthesized. The Rule-Based Module is developed to compensate for the actuator delay and inconsistency in the fill-phase and torque-phase of a shift. The Sliding Control Module is developed to compensate for uncertainties during the speed-phase operation. In this research, two types of output signals are used, namely, output torque and output acceleration. Therefore, a torque-based controller and an acceleration-based controller are developed. (3) Discrete Multiple Surface Sliding Controller is developed to control systems whose control input is not explicitly related to the control output. (4) To estimate the output torque for the output-torque-based Hybrid algorithm, a Neural Network estimator is developed to identify the transmission input/output torques and the clutch frictional coefficients. Promising results are shown through computer simulation. Limitations of this recurrent network approach are also identified. (5) Two novel acceleration estimation schemes are developed. It is proved that the proposed estimation schemes can improve the performance upon any base-line estimation from available schemes. (6) It is shown that the proposed estimation schemes provide good estimation of acceleration for the Hybrid approach. It is also illustrated that the output-acceleration-based Hybrid controller can achieve more robust performance compared to two of the most recently developed transmission clutch-to-clutch control schemes.