Study On Simulation Platform For Automotive Electronic Control System

To develop the advanced Automotive Electronic Control System, the interactions with road environment as well as other traffic vehicles need to be studied. However, the conventional math tools were mainly designed to simulate driver-commanded vehicle dynamics, with limited capability on modeling of various environmental and traffic condition, and their interaction to drive and vehicle controls, as in features of adaptive cruise control, collision avoidance, and lane change. Thus, the development of Next Generation Vehicle Simulation Environment is a new issue in the Vehicle Simulation Field, ant it is also the Motivation and Academic Values of this thesis.Next Generation Vehicle Simulation Environment should solve many issues of key subsystems, such as high-precision real-time vehicle dynamics model, sensor model, road-network, driving scenario and driver model. These issues include modeling issues, integration issues and numerical solving issues. For these issues, carried out some analysis and proposed some engineering solutions.Although traditional multi-body vehicle dynamics model can make a full description of the movement of each subsystem and the interaction between them, but has a low computational efficiency to handle Stiff Differential Equations, so it doesn't meet the requirements of real time simulation. To handle this issue, proposed a method that the suspension subsystems are modeled as mass-less composite joints and the constraint equations of those joints are formulated based on Kane's method. A vehicle dynamics model was created based-on the method to reduce the computation time of simulation without degradation of kinematical accuracy.Sensor technology plays a critical role in the Automotive Electronic Control System, but traditional vehicle simulation environment lacks of effective sensor simulations. By analyzing the functions of the car sensor, the sensor models presented here are part of the functional sensor hierarchy, incorporating geometric models, abstract noise characteristics. Furthermore, describes the IO structures, data structures and interfaces of these functional sensor models. Focus on the radar sensor and vision sensor function model and theirs IO structures and data structures, and study the establishment of the two models:Real-time collision detection, realistic image generation, and Gaussian white noise generation.In the traditional vehicle simulation environment, the establishment of roads and driving scenarios are based on several data table describes, inefficient and error prone. Have a research on developing an interactive graphical editor to build the main road environments and driving scenarios, and focus on the issues and main method using this editor to generate realistic driving scenarios, and in-depth study of the local traffic environment generation methods. Road environments and driving scenarios are displayed in the 3D animator, developing the 3D animator based-on a graphics engine named as OSG.A typical simulation is composed of a widely and large amount models, such as the road environment, driving scenario, vehicle model, control model and sensor model. For the effective management of these models, have a research on the object-oriented simulation database. In order to reuse existing models and reduce the modeling errors caused by manual integration, focus on the simulink model file automatically generated method and model integration based on this method. Furthermore, to meet the engineering requirements of distributed simulation and to support models plug and play, develop a distributed simulation model framework.Major Innovations of the Dissertation:1. Build a real-time multi-body vehicle dynamics model based on suspension composite joints and Kane's method to meet the requirements of precision and real-time for Automotive Electronic Control System.2. For the difficulties of establishment of a complete physical sensor model, proposed the concept of the sensor functional model.and had a research on the key issues and methods for radar and vision sensor functional model. 3. Established an interactive and graphical environment to build road environments and driving senarios. Developed an interactive real-time 3D animator, which can generate a executable program with the scene database4. Developed a Simulink MDL file automatic generator to support simulink based subsystem model integration and simulation.Developed a framework to support distributed simulation.