摘要
在汽车防抱死制动系统(ABS)中,压力调节器和轮速传感器起着非常重要的作用,为了进一步完善汽车防抱死制动系统的制动性能,文中提出一种基于概率神经网络(PNN)的压力调节器和轮速传感器的故障诊断方法;基于高附着均一路面,起车时制动及单一的压力调节器或者轮速传感器故障的试验数据,分别建立了基于概率神经网络的压力调节器故障诊断模型和轮速传感器故障诊断模型,并与BP神经网络进行了比较;仿真结果表明,利用相同的训练样本集对概率神经网络和BP神经网络进行训练时,基于概率神经网络的压力调节器故障诊断模型和轮速传感器故障诊断模型在训练时间和诊断精度上明显优于BP神经网络,并且利用测试样本对建好的压力调节器故障模型和轮速传感器故障模型进行检测时,无论测试样本的顺序发生什么变化,基于概率神经网络的故障模型都能够准确的进行故障识别。
Pressure regulator and wheel speed sensor play an important role in the anti-lock braking system of automobiles.In order to further improve the braking performance of the anti-lock braking system of automobiles,a fault diagnosis method of pressure regulator and wheel speed sensor based on probabilistic neural network is proposed.Based on the test data of braking and single pressure regulator or wheel speed sensor faults when start-up on high adhesion uniform road surface,the fault diagnosis models of pressure regulator and wheel speed sensor based on probabilistic neural network are established respectively,and compared with BP neural network.The simulation results show that when the probabilistic neural network and BP neural network are trained with the same training sample set,the fault diagnosis model of pressure regulator and wheel speed sensor based on probabilistic neural network is obviously superior to BP neural network in training time and diagnostic accuracy,and when the fault models of pressure regulator and wheel speed sensor are detected by the test sample,the fault models based on probabilistic neural network can accurately identify the faults no matter what the order of test samples changes.
作者
孙丽娜
王佳庆
黄永红
Sun Lina;Wang Jiaqing;Huang Yonghong(Mechatronics Engineering Department,Suzhou Industrial Park Institute of Vocational Technology,Suzhou 215123,China;School of Electrical and Information Engineering,Jiangsu University,Zhenjiang 212013,China)
出处
《计算机测量与控制》
2020年第4期16-21,共6页
Computer Measurement &Control
基金
江苏省自然科学基金面上项目(BK20151345)。
关键词
防抱死制动系统
压力调节器
轮速传感器
概率神经网络
anti-lock braking system
pressure regulator
wheel speed sensor
probabilistic neural network
