Study On Control Of Automated Mechanical Transmission Based On Driver’s Starting Intentions

As an increasing application of automatic transmissions, automatedmechanical transmission (AMT) adding an inexpensive control unit toclassical manual transmissions is considered as a mean of reducing fuelcost, easing driver task, and thus meeting driver’s expectations. Massiveexperiments have confirmed the feasibility of AMT, which is a focalpoint in domestic automatic transmission research and development andis worthy to be industrialized. However, control of vehicle launch isalways the emphasis and difficulty for the control of AMT.A dynamic model for the driveline of AMT is proposed on the basis ofengine model, clutch model, vehicle model and actuator model. Thepresented model can provide platform for the simulation and control ofAMT system.The research interest is focused on a driver’s starting intentions inorder to make AMT system recognize different kinds of startingintentions. An orthogonal test design is applied to treat the variables andtheir combinations for the driver’s decision making during launching thevehicle. The influences of the pedal opening, the first derivative of thepedal opening, and the second derivative of the pedal opening arequantified. According to these factors, a hidden Markov model (HMM)is used to model and recognize three different kinds of starting intentions,including fast starting, mild starting and slow starting.This paper explores association between driver’s starting intentionsand control of dry clutch engagement during vehicle launch in order todevelop an intelligent automated mechanical transmission system. Theoptimal engagement laws are deduced by finding a compromise among friction loss, shock intensity, engine torque and engine angularacceleration, based upon the extremum value theorem. The pedalopening, the first derivative of the pedal opening and the secondderivative of the pedal opening are chosen as the input while the shockintensity as the output in fuzzy control. According to the optimal engineangular speed and engine torque, the optimal accelerator opening isobtained from engine steady output torque model. The optimalaccelerator opening reflects driver’s starting intentions correctly andsatisfies the optimal control of clutch engagement.Clutch slipping can be evaluated by slipping work, but the wearduring slipping can not be quantified. Thus, this paper explores wearmechanism and dynamic optimization of the dry clutch during slipping.A wear model for wear evaluation during the steady stage of wear isproposed based on fractal theory, adhesive wear theory andMajumdar-Bhushan Fractal Contact Model. The expression of the rate ofwear depth for each engagement during the optimal control of clutchengagement is deduced in order to realize the dynamic optimization.In order to verify the accuracy and adaptability of the startingintentions models for different kinds of drivers, an experimental studyon the starting intentions is carried out by the hardware CANcaseXL andthe software CANape. The experiment interest is focused on datacollection of accelerator opening and vehicle speed of different kinds ofdrivers. The results show that the HMM model has good accuracy andadaptability for the starting intentions of different kinds of drivers.The study presented in this paper plays an important role in the keytechniques during vehicle launch of AMT, and it is significant topromote the industrialization of AMT.