Study On Personalized Control Strategy Of Full X-by-wire Electric Vehicle Based On Driving Style

Tchnology breakthrough and leading concept becomes the dual engine of automobile industry innovation.The full x-by-wire electric vehicle is considered to represent the trend of the future electric vehicle chassis because of its high integration of advanced technologies such as four-wheel independent steering,four-wheel independent drive and four-wheel independent braking.The driving system of the platform can control the characteristics of free configuration,which provides opportunities for the optimization of dynamic characteristics under various concepts.Steering characteristics,as a key factor affecting the handling and stability performance,also largely determine the driver’s driving comfort.Based on the driving style,the research on the personalized design and relevant control strategies of the steering characteristics for the full x-by-wire electric vehicle is the embodiment and implementation of the concept of "vehicle adapts to people" in the vehicle intelligent development system.Based on the National Natural Science Foundation Project "Research on control mechanism and evaluation method of new type of steer-by-wire system based on driver’s characteristics"(No.51575223),this paper starts from the driving style as the personalized benchmark,designs the implicit personalized steering characteristics of full x-by-wire electric vehicle,and develops the driving system with layered integrated control to provide handling stability guarantee,in order to achieve the goal of "safety + comfort";At the same time,for the fuzzy concept of driving style and the coupling problem of multiple benchmarks in this process,the paper puts forward the extended definition of driving style,and driver’s return is defined as a general object in the field of philosophy,using object three-level structure to establish driving style conceptualframework involving multi driver terms,so as to realize to uniformly explain those existing personalized benchmarks and guide the driving style recognition and personalized design.The specific work of this paper includes:1)The personalized design of steering characteristics based on the driving style calls for understanding the connotation of driving style,so aiming at the problem of the fuzzy concept of driving style and the coupling of multiple benchmarks,this paper analyzes the common description of the existing definitions,and puts forward the extended definition of driving style from the internal mechanism and external performance,so as to clarify the relationship between driving skill,driving preference,behavior characteristics,driving characteristic and driving habit;The driver’s return is defined as a general object in the field of philosophy,using the threelevel structure of the object,establish a conceptual framework of driving style involving multi driver terms,display the connotation and relationship of these terms in a more three-dimensional way with structural description,and use mathematical description to sort out and form driving style application explanation including personalization;According to the "strengthening" mechanism,explore the formation process of driving style and improve the conceptual framework;For the feature layer,in a human-vehicle-environment interaction system with multi-levels and multi-links,interpreting the behavior mechanism of driving style according to behavior time and behavior space;The conceptual framework of driving style is integrated with the existing recognition methods to form three recognition ideas,and the objective recognition methods based on quantitative indicators are sorted out;finally,the exiting theory and the personalized application about the driving style obtains uniform interpretation,so as to guide driving style recognition and personalized design.2)The exiting synergistically matching trendency between the driving style and the vehicle dynamic characteristics are obtained,under the conceptual framework of driving style,based on the explicit connotation of driving style,a feasibility explanation about the mechanism and theory for this matching is proposed;and the tactical decision-making style of tactical decisionmaking behavior link in cuvre is the key for the steering characteristics matching;This parper calls it effective steering driving mode,and proposes the overall ecognition scheme of tactical decision-making style in the effective steering driving mode,based on the driving stylerecognition system described in above;The driving simulator is used to carry out the multi curve natural driving experiment of the driver’s sample to establish the driving style information base;For the tactical decision-making style under the single effective steering driving mode,proposing a multi-dimensional describing schema for feature space with {working condition dimension,trade-off dimension,fluctuation dimension},and according to it,the the index group of {turning radius,average vehicle speed,vehicle speed fluctuation amplitude} is built;And it represents a feature point,obtaining all feature points of all driver’s samples can form feature set for training and testing of driving style recognition.The kmeans algorithm is used to pre-classify the feature points in the dense area of working condition level,and to label the driver’s samples,so as to build the driving style recognition model based on the decision tree algorithm by semisupervised manner,and its outputs of digital code will be used by the driving system of the full x-by-wire electric vehicle.Based on the driving simulator data and real vehicle data,the accuracy,generalization ability and robustness of driving style recognition model are verified.3)Based on the flexible layout of the chassis of the full x-by-wire electric vehicle and the characteristics of extremely high controllable freedom,a driving system with 5-layer integrated control and multi-mode adaptive steering characteristics is developed to improve the driving comfort and ensure the handling stability at the same time: in addition to identifying the basic parameters,the identification and identification layer is also responsible for the identification of driving style and steering intention;In the mode configuration layer,a motion reference model reflecting the driver’s needs is arranged,and the yaw rate reference model is divided into two parts: steering gain and yaw rate gain.The former adapts driving style to form a long-term mode,while the latter adapts steering intention to form a short-term mode.The smooth switching of long-term and short-term modes is realized by moving average window and cosine function respectively.In the motion control layer,the sliding mode algorithm is used to build the nonlinear motion controller,which is used to calculate the virtual target force / torque of the center of mass according to the target motion output from the upper layer.When adjusting the controller parameters,the "total disturbance" is fully considered to improve the system robustness.In the force distribution layer,the influence of actuator saturation constraint and load transfer is fully considered,and the calculation load and accuracy are considered.Octagon isselected to fit the friction circle linearly,and the tire force loss area is arranged in the direction that has little influence on the vehicle motion control.The objective function is defined as the maximum stability margin and the minimum longitudinal force error of the four-wheel tire,and the multi mode is guaranteed with sufficient lateral force.Finally,the force distribution layer is constructed as a quadratic programming problem with two linear equality constraints and 40 linear inequality constraints.The executive layer uses the inverse model of the arctangent function tire model to convert the virtual target tire force into the executable command of the actuator,that is,the driving torque of the hub motor and the steering angle of the steering motor.The joint simulation based on Car Sim and MATLAB / Simulink verifies the effectiveness of each mode and the handling stability of the whole vehicle under various handling conditions.Finally,the subjective evaluation experiment from the sports driver and the soothing driver by the driving simulator is carried out.The results show that the long-term mode gradually adapts to the steering mode matching the driving style type and is stable and smooth.Under the cross evaluation way,the two drivers also give a relatively satisfactory rating of the adaptive effect of the steering characteristics.The main innovation points of this dissertation are followed:1)Propose the extended definition of driving style from two perspectives of internal mechanism and external performance,and establish the conceptual framework of driving style with three-layers,so as to realize to uniformly explain those existing personalized benchmarks.The personalized design of steering characteristics based on the driving style calls for understanding the connotation of driving style,so in view of the current situation of the coupling application of fuzzy concept and multi benchmark of driving style,the extended definition of driving style is proposed from two perspectives of internal mechanism and external performance.According to the three-layers structure of the general object in the field of philosophy,a conceptual framework of driving style is established to clarify the relationship between driving skills,driving preferences,behavioral characteristics,driving characteristics and driving habits.In this framework,the application interpretation,formation interpretation and behavioral mechanism interpretation of driving style are carried out.Then,it integrates with the existing recognition methods to make up a recognition idea interpretation,so as to guide thefollowing driving style recognition and personalized design.2)Aiming at the tactical decision-making style in the effective steering driving mode,bulid the index group of {turning radius,average vehicle speed,vehicle speed fluctuation amplitude}and a driving style recognition model based on decision tree algorithm.The exiting synergistically matching trendency between the driving style and the vehicle dynamic characteristics are obtained,and based on the explicit connotation of driving style,a feasibility explanation about the mechanism and theory for this matching is proposed;The tactical decision-making style of tactical decision-making behavior link in cuvre is the key for the steering characteristics matching,which is called as effective steering driving mode;Aiming at the problem that the exiting index group lack of descriptive ability for this mode,this paper proposes a multi-dimensional describing schema for feature space with {working condition dimension,trade-off dimension,fluctuation dimension},and according to it,builds the index group of {turning radius,average vehicle speed,vehicle speed fluctuation amplitude},constructing feature set.Kmeans algorithm is used to pre classify and label the feature points in the dense area of working condition level,so as to build a driving style recognition model based on decision tree algorithm.Its outputs of digital code will be used by the driving system of the full x-by-wire electric vehicle.3)Aiming at the full x-by-wire electric vehicle,develop a driving system with 5-layer integrated control and multi-mode adaptive steering characteristics,and the long-term mode is formed by the steering gain adaptive driving style,then the short-term mode is formed by the yaw rate gain adaptive steering intention,improving the driving comfort and ensuring the handling stability at the same time.Based on the goal of implicit personalized design and handling stability guarantee of steering characteristics,combined with the characteristics of full x-by-wire electric vehicle chassis,a driving system with 5-layer integrated control and multimode adaptive steering characteristics is developed: the long-term mode is formed by the steering gain adaptive driving style,the short-term mode is formed by the yaw rate gain adaptive steering intention,the smooth switching of long-term and short-term modes is realized by moving average window and cosine function respectively;the nonlinear sliding mode algorithm is combined with the dynamic potential of the full x-by-wire electric vehicle,and each mode isprovided with handling stability guarantee,for achieving the goal of "safety + comfort".