Data-driven Predictive Control Of Gearshifts And FPGA Implementation For Dual-clutch Transmissions

Dual-Clutch Transmissions (DCT) was developed from the basement of ManualTransmission (MT),it inherit MT’s high transmission efciency and can shift withoutpower interrupt as Automatic Transmission (AT) work as.The cars which fitted withDCT have the same driving convenience and comfort like AT, and have less power losingand high fuel economy.For similar structure MT and DCT they have, MT can be easychanged to modification rapid production of DCT. DCT’s compact structure, small vol-ume are assembled in the car needs to be smaller space, small vehicle due to limited space,more suitable for assembly of DCT. Therefore, in recent years DCT has become the majorcar companies invest a lot of manpower material resources research and development ofnew hot spots, nissan, mitsubishi, BMW and porsche all have developed their own DCTproducts.the shift process of DCT can be called shifting clutch switch type, can be dividedinto two stages torque phase and inertia phase, phase torque redistribution happen intorque phase which transmission ratio without drastic changes,clutch synchronization andslipping happen in inertial phase with speed and transmission ratio of dramatic change, itis easy to generate large shift shock during inertia phase.if slipping time is too long thatit also can produce higher sliding friction work, those afect the service life of clutch. Soit is very needed a closed-loop control during inertia phase in order to synchronize theclutch slipping smoothly and make the transmission of output torque change is relativelystable, reducing the impact due to the torque caused by the volatility of shift. Shifttime, shifting jerk and slipping work is contradictory evaluation index, how to balanceand achieve the optimal shift control. Because the DCT lack of one-way clutch whichsimilar to AT hydraulic torque converter, if the shift control is not good, there will beinterference between the clutch torque and output torque.Key technology of wet DCT shift process is studied in this paper, the whole contentis divided into the following five parts: The first chapter is mainly for the introduction. The research background of thisarticle has carried on the simple introduction and summarizes the history of DCT auto-matic transmission development and research status in China and abroad. some DCTshift control results are analyzed and summarized and put forward in this paper.Chapter2establish DCT’s model of shifting vehicle system model process,in order todo the experiment of simulation. Established model include: engine model, and hydraulicdrive, dual clutch transmissions and synchronizer model, load model and dynamic modelof the vehicle. In this chapter,a kind of car with a wet dual clutch automatic transmissionwas choosed as the modeling object.the whole simulation model was developed by thesoftware of AMESim which can combine with Matlab/Simulink simulation,the simulationmodel was prepared for the third chapter as simulation platform.Chapter3the controller for DCT inertial phase shift is designed using data-drivenpredictive control method. the method can be obtained the future output prediction equa-tion from simulation data of system model input and output data. Choosing the systemoutput speed diference of clutch, throttle opening and engine proportional solenoid valvefor the input data. Shifting strategies use clutch speed trajectory tracking method to syn-chronize clutch speed diference tracking on predefined target reference trajectory at thesame time by controlling the engine throttle opening,then engine speed change cooperat-ing with shifting clutch of inertial phase synchronization. The Restrictions on the wholevehicle system is mechanical, so the input and output is limited by certain constraints.thecontrol problem is converted to constraint optimization problems which can be appliedto QP method to solve constrained optimization problem to.then the final control outputare getted. The last will be the shift controller designed by MATLAB/Simulink to real-ize, and the second chapter AMESim simulation model is set up by the joint simulationvalidates the efectiveness of the controller.Chapter4is for the FPGA implementation of the controller for DCT shifting.Data-driven predictive control method and an online optimization method involved.as the con-straints of system cause the controller work slow and output prediction equation involvesa lot of matrix calculation,so the speed of solving controller became become the primaryproblem of engineering application. the FPGA has powerful computing capacity and canmeet the realization of the function of the complex complex algorithm, in order to makefull use of FPGA’s hardware characteristics of parallel computing to accelerate the oper-ation of the algorithm. For the convenience of controller design of the hardware circuitimplementation, here use advanced synthesization tools Vivado HDL whic as FPGA de-velopment tool for the realization of a controller hardware. The Final implementation ofshifting controller is downloaded to the FPGA,and be checked with the prbuild simula- tion platform of dSPACE real-time simulation system for rapid prototyping experiment.Experiments show that this design of shifting controller meet the demand of real-timeand control.