中文摘要：

为了提高汽车ABS整车检测效率,开发了一种新型的基于台架的汽车ABS整车检测系统。该系统利用飞轮的转动惯量模拟车辆在道路上运动的平动惯量;通过扭矩控制器在4个车轮所在滚筒上加载不同的扭矩,实现不同路面附着系数的动态模拟;采用基于CAN总线的分布式网络测控技术完成台架的运动控制和车辆速度数据的采集;利用BP神经网络自学习功能分析台架检测数据,总结大量模式映射关系,用训练好的网络实现检测结果自动分类。室内台架检测结果和道路试验结果对比分析表明：台架检测结果与道路试验结果基本相同,主要参数误差小于4%,因此,台架检测系统能够准确地反映装有ABS的车辆在不同路面工况下的制动性能。

英文摘要：

In order to improve the detection efficiency of automobile ABS,a novel detection system was developed based on bench.The inertia of moving automobile was simulated by using the rotational inertia of flywheel.Four torque controllers were adopted to load different torques on four rollers supporting four wheels of automobile,so that different road adhesion coefficients were simulated.CAN-bus-based distributed network control technology was used to complete the motion control of bench and the data acquisition of automotive speed.BP neural network was used to analyze the test data for its self-learning function.The summarized mapping relationship of ABS work states was stored in the network,and the automatic classification of test results was achieved by using the network.Comparative analysis of bench test result and road experiment result for ABS shows that the results are basically same,the errors of main parameters are less than 4%,so bench test can more accurately reflect the brake performances of automobile equipped with ABS in different road environments.3 tabs,10 figs,20 refs.