Research On Vehicle ABS Test-bed Control System And Test Analysis

Vehicle ABS test-bed, developed for vehicle indoor bench test, is a complex inspection device that consists of electromechanical and pneumatic components. It has requirements for arduous tasks, complicated operation procedures and high safety and so on. To ensure the ABS test-bed working in good condition with reliability and accuracy, the thesis aims to develop the computer measurement & control system based on CAN. Besides that, multi-vehicle comparison tests and data analysis work are carried out based on the test-bed.On the base of studying the vehicle ABS theory and its test methods, an ABS test-bed measurement & control system is developed using CAN bus. It uses the distributed CAN system to distribute complicated and heavy measurement & control tasks to each smart module. Each of the smart modules has independent functions with measurement, control, calibration and self-diagnosis and so on. PC sends messages to the smart modules as co-ordinatior so as to ensure the reliability and real-time data transmission. The measurement & control system uses PWM technique to control the input currents of four torque controllers to produce different torques, which can realize the simulation of different road adhesion coefficients. Travelling control system is composed of a transducer, an AC servo motor, an optical encoder and other components, which achieves precise control of the movable bench's position. The controlling system adopts CPLD technique to realize the high accuracy and real-time speed acquisition of the four wheels and body of vehicles. Finally, bench experiments are carried out with different vehicles, different road adhesion coefficients, and the experiment data is analyzed. The thesis also includes an ABS fault classification model based on BP neural network using a large number of experimental data to train the network, and then realizes the automatic evaluation of test results.The result shows that the ABS test-bed controlling system can collect data of the vehicle and the wheels speed accurately in real-time when the vehicles brake emergently at high speed. The trained neural network can divide the test results into three categories of ABS normal, ABS fault and insufficient braking force.