Research On Adhesion Control Method Based On Model Predictive Control

With the ever-increasing demand and large market,electric multiple units(EMU)are favored by people due to their advantages such as large volume,high speed and high punctuality.At the same time,they also arouse general concern about train safety.When the train is running,the traction/braking force generated by the motor is transformed by the adhesion force between the wheels and rails.If the traction force increases beyond the maximum adhesion between the wheels and rails,the wheel could be idling/slide easily.Therefore,this paper designs an adhesion control method based on model predictive control,which can not only suppress the wheelset idling in time,but also control the train to run stably near the adhesion peak point to achieve the maximum adhesion utilization.First of all,considering that each adhesion curve has an adhesion peak point,therefore,an optimal slip velocity searching method is designed,which can judge the running status of the train through the estimated adhesion coefficient and slip velocity,and then adjust and output the reference slip velocity in real time.Finally,the reference slip velocity can stabilize at the optimal slip velocity of the corresponding adhesion characteristic curve.Then,a dynamic model of train is built on MATLAB/Simulink software,and a torque controller for adhesion control is designed to verify the effectiveness of the searching method for the optimal slip velocity.Secondly,the nonlinear model predictive controller based on the nonlinear train dynamics model will cause difficulties in optimization,which is not able to satisfy the real-time performance of the actual system.At the same time,the main nonlinear characteristic of the train is concentrated on the adhesion characteristics between the wheel and rail.Therefore,the T-S multi-model modeling method is adopted to analyze the adhesion characteristics.The clustering method based on local linearity is used to cluster the off-line sample data of adhesion characteristics curve,to determine the number of clusters and the corresponding cluster centers.When the antecedent parameters of the T-S fuzzy rule are obtained,the least square method is used to calculate the local linearity parameters.Finally,based on the MATLAB/Simulink software,the adhesion characteristics under the dry and wet rails are modeled separately,and the modeling results are analyzed to verify the effectiveness of the T-S multi-model modeling method.Finally,in order to realize the best utilization of adhesion,an adhesion controller based on model predictive control is designed.The reference input of the predictive controller is the optimal slip velocity output by the optimal slip velocity searching module,and the predictive model is the optimized multi-model.In the design of the predictive controller,the objective function and the constraints of the traction system are used for rolling optimization,and then the optimal traction torque at the current moment is obtained to realize the stable tracking of the slip velocity to the optimal slip velocity and to achieve the adhesion utilization.A semi-physical platform of the adhesion control system is built based on the PLECS software and the RT BOX platform,and the switching rail conditions is simulated to verify the effectiveness of the designed adhesion control method based on model predictive control method.