SOPC Compound Controller For Automobile Active Suspension

Active suspension can significantly enhance their performance in the limousine and specialty vehicles, spontaneously the application has become inevitable. Current research on active suspension control law maily focused on the ride, little literature on the comfort and trafficability, moreover, the reseach of muti-control or predictive control and its hardware implementation is rarely appear. This thesis presents complete solutions of multi-controllor for air-adjustable active suspension, including software and hardware design, which combined with barriers-predictived control and attitude control.This paper analyzes the characteristics of the air adjustable suspension, using its initiative to adjust the high frequency response and the speed of passive stretching proposed combination of active and passive strategy when adopt obstacle. In order to predicted obstacle and attitude adjustment, here put forward a multi-targeted compound control scheme, control strategy includes the following four parts:1. According to the displacement and acceleration of suspension controllor would determined the road conditions emergency and vehicles conditions, developed fuzzy rules to alter suspension damping; 2. Select the body acceleration, suspension and tires displacement, to build an objective function, so as to obtained optimal control law by LQG; 3. Knowing the distance and height of road barriers, using the Gauss function, predictive control the body height and suspension force to expand the buffer zone in order to adopt the obstacle smoothly; 4. Combined with changes in damping, a called adaptive PID attitude controlling would maintain vehicle stability, then it comprehensive LQG optimal control law and Gauss obstacle adopting to eventually alter the suspension force. What is more, Matlab/Simulink was used to established models all of control algorithms and the road input both of the normal one and the convex one, and auto 2-DOF1/4 and 7-DOF full suspension models, as well as and assemble model of scopes. Finally, using the models above to simulate the crucial performance of the compound control system, we had made amendment repeatly to complete the algorithm-level design and implementation.This paper also studys depthly on measurements of body attidute. Here proposed a streamline measurement method to work out the angles, which using three-axis acceleration sensor combimed with rear wheel speed encoders and for its limitations of computational complexity and modeldependency, an optimization by using the neural network mapping also proposed. The latter arithmetic was based on the stationarity of automobile movement. While clustering the harmful section of inertial, and using high-precision gyro output for line training, the neural network neural due with non-deterministic multiple input multiple output mapping, to reveal its nonlinear function learning and generalization capabilities. Modeling and simulation at Matlab show the effectiveness and reliability of this algorithm.Finally, from the perspective of high-speed hardware implementation, here analyzed the various characteristics of the algorithms, then using the Xilinx FPGA platform to build this system. The research focued on optimize of architecture and module resource distributing. This paper discusses the relationship between complex algorithm time costing on implementation while through hardware and software. According to algorithm characteristics, to make it simple, we rational choice top-down hardware and software co-design from system-level and RTL-level. System-level design changes the Simulink model using the algorithm Middleware System-Generator into Xilinx ISE project file. RLT-level design directly descripte the module by Verilog HDL in the ISE environment, and then integrated, simulation, layout, as well as generate an executable file for the board debugging. Hardware simulations show that the use of reconfigurable parallel processing logic to obtain a high performance and low overhead of hardware.