Research On The Method Of Formulating DCT Shift Strategy For Straight Roads Based On Data Mining

The research and development of automatic transmission is an important part in the development process of automobile industry,which represents the advanced development level of automobile transmission system,and the formulation of shift strategy is the core research content of transmission theory.The quality of shift strategy directly affects the comprehensive performance of vehicles and determines the shift quality of vehicles.At present,the formulation method of gear shifting strategy mostly depends on manual calibration and theoretical optimization,which will have a big gap with the effect of shifting gears on real roads.In order to achieve the goal of achieving the same vehicle performance control effect between the shift process of the driver driving the automatic transmission vehicle and the shift operation of the excellent driver driving the manual transmission vehicle,this paper combines the data mining technology with the formulation of the shift strategy,and makes full use of the driver’s historical manipulation data to mine its hidden shift strategy.Taking the driving experimental data collected by skilled drivers driving manual transmission vehicles as data mining objects,the research on shifting strategy mining under constant speed and accelerated driving intention is carried out.The main research contents are as follows:(1)Pretreatment and cleaning of vehicle driving test data.Due to the actual project requirements and limitations,the driving test data of manual transmission vehicles cannot be provided,so the driving data collected by fuel vehicles equipped with seven-speed DCT will be used instead of the test data of manual transmission vehicles for the time being.This paper introduces the data acquisition equipment,experimental method and driving route in driving experiment,and resamples the problem of data asynchronization between different characteristic parameters Cleaning driving characteristic data from two aspects of singular value detection and elimination and wavelet denoising respectively;Based on the sensor principle,the current road slope is estimated by using the vehicle longitudinal acceleration and measured acceleration parameters,and the steering wheel angle range is limited to eliminate corners After the above data processing work,the highquality driving data of the straight road is finally obtained.(2)Solving the optimal economic and dynamic shift law based on driving data.Firstly,aiming at the problem that some characteristic data values of engine bench test are quite different from the measured values of driving test,the causes of the above problems are analyzed Four characteristic parameters of engine speed,engine torque,fuel consumption and throttle opening are extracted from driving data,and the value range is set.The model of engine fuel consumption and torque is constructed by interpolation fitting,and the missing parts in the model are supplemented by quartic polynomial fitting Smoothing the obtained engine model surface,using the model data,choosing analytical method to solve the optimal economic and dynamic shift rules as the comparison benchmark of the shift rules for subsequent real vehicle mining.(3)Mining research on two-parameter shift strategy under constant speed condition.Firstly,the acceleration threshold is set to divide the driving data into three categories:uniform speed,acceleration and deceleration;Make a two-parameter shift rule for uniform working condition,extract the speed and throttle opening characteristic parameters of each gear in uniform working condition data,select DBSCAN algorithm to detect and eliminate outliers in two-dimensional data of each gear,use Alpha-shape algorithm to detect the boundary points of data clusters respectively for the processed seven groups of gear data,and then interpolate the shift boundary points into the boundary lines of lifting gears;Compare the excavated shift strategy with the optimal economic shift law to verify the true effectiveness of the excavated shift strategy;For the shift speed corresponding to the missing part of throttle opening data,select the shift time corresponding to the best economic shift strategy to fill it;(4)Mining research on three-parameter shift strategy under acceleration condition.For the acceleration condition,the three-parameter shift strategy is made by selecting the characteristic parameters of vehicle speed,throttle opening and acceleration,and the three-dimensional data of each gear is extracted and the outliers of the three-dimensional data are eliminated by using OCSVM algorithm.The union set of the points of the twodimensional detection data sets is taken as the three-dimensional data boundary point,and the shifting surface of the acceleration condition is obtained by surface interpolation of the lifting gear boundary point by cubic method;Compare the shift curves of different accelerations with the standard shift curves to verify their accuracy.To solve the problem of incomplete shift surface caused by incomplete data,the nonlinear least square method is used to fit the shift surface to obtain a complete three-parameter shift strategy.