Research On Self-Learning Algorithm Of AMT Clutch Position For Electric Vehicle Without Power Interruption

With global warming,resource depletion and other increasingly serious problems,the automobile industry has undergone major changes,and pure electric vehicles have become the mainstream of development.Electric drive system is one of the core components of electric vehicles.At present,most electric vehicles in the market are only matched with fixed speed reducer,and equipped with two-speed gearbox,which can make the drive motor always run in a high efficiency range and improve the power and economy of the whole vehicle.During the working process of the gearbox,the clutch moves from the maximum displacement to the minimum displacement,and the actual working range is the distance from just not transmitting torque to just transmitting rated torque.Therefore,the identification of these two positions is very important.Moreover,when the car runs for a certain mileage,the clutch actuator and the clutch friction plate are worn or the diaphragm spring is tired,the initial contact position of the clutch and the just-completely-combined position will shift,so it is necessary to add the self-learning function to accurately identify the clutch friction point and the completely-combined point,so as to make up for the influence of clutch wear on the gear shifting process.In this paper,I-AMT（Inverse AMT）,a new type of unpowered two-speed transmission for electric vehicles,is taken as the research object.Aiming at the problem that the positions of the sliding friction point and the full engagement point change after the friction plate clutch is worn,and the actual positions of the first and second gears shift,a self-learning strategy for the sliding friction point and the full engagement point of the clutch is proposed.Through Catia modeling of clutch actuator,motion force analysis,Simulink vehicle modeling simulation,bench test and real vehicle test,the self-learning strategy of clutch friction point and perfect joint point is studied.Simulation and experimental results show that the self-learning strategy can control the clutch to track the target curve stably and accurately capture the position of friction point and perfect joint point,and make up for the position deviation of the first and second gears caused by clutch wear.The specific work of this paper is as follows:1.Introduction and theoretical analysis.Firstly,the structure of two-speed gearbox without power interruption and the working principle of overrunning clutch are introduced,and the power transmission route in two forward gears and reverse gears is analyzed.The shift process is analyzed,and the principle of upshift without power interruption is explained.The wet friction plate clutch and its actuator are introduced in detail,and their motion is analyzed.2.A clutch self-learning strategy is put forward to solve the problems of the position change of sliding friction point and perfect joint point and the actual position deviation of the first and second gears after the wet friction plate clutch is worn.At the same time,in order to realize the accurate and slow control of clutch displacement,a feedforward and feedback controller is designed.3.Build the whole vehicle model,clutch actuator model,drive motor model,feedforward-feedback controller and so on with Simulink.Aiming at the characteristic that the simulation model can detect the clutch torque in real time,the self-learning strategy is slightly simplified,and the effectiveness of the self-learning strategy and the control effect of the feedforward and feedback controller are verified in the simulation model.4.The test bench and the real vehicle test platform were built,and the hardware design and the bottom driver program of the transmission control unit TCU were written.The control layer model of TCU was built by using the model-based design technology（MBD）,and the control layer model was generated by using the automatic code generation technology and burned into the TCU together with the bottom program for bench and real vehicle test.This paper introduces the hardware composition of the Internet of Vehicles development board,rewrites the underlying driver and writes the server program.Finally,the self-learning results are uploaded to the remote server through the Internet of Vehicles development board.The test results show that the selflearning strategy adopted in this paper is still reliable and effective on the test bench and real vehicle,and the success rate is high,which can effectively compensate the shift of working point caused by the wear of I-AMT clutch,thus ensuring the comprehensive performance of shifting.