基于BP神经网络的EHB主缸液压力估计  

BP Neural Network-Based Master Cylinder Pressure Estimation for EHB

作　　者：史彪飞 王磊[2] 梁海强 李荣利 梁超 Shi Biaofei;Wang Lei;Liang Haiqiang;Li Rongli;Liang Chao(Tsinghua University,Beijing 100084;Beijing Automotive Technology Center,Beijing 101300)

机构地区：[1]清华大学,北京100084 [2]北京汽车研究总院有限公司,北京101300

出　　处：《汽车技术》2025年第1期57-62,共6页Automobile Technology

基　　金：山东省重点研发计划(2023CXGC010214)。

摘　　要：电子液压制动(EHB)系统主缸液压力估计对降低EHB的传感器依赖性至关重要,基于BP神经网络进行主缸液压力估计。首先开展了实车道路试验,并采集车速、主缸活塞位移、主缸活塞速度和主缸液压力等数据。然后,以主缸活塞位移和主缸活塞速度为特征输入、以实际主缸液压力为目标输出建立BP神经网络,并采用训练集数据及梯度下降算法对BP神经网络进行训练。最后,利用测试集数据对液压力估计效果进行验证。结果表明,所提算法比基于动态位移压力模型和基于LSTM的估计算法估计误差分别减小38%和15%。The master cylinder pressure estimation of the Electro-Hydraulic Brake(EHB)system is crucial to reduce the sensor dependence of EHB.In this paper,the master cylinder pressure is estimated based on BP neural network.First,a real-vehicle road test is carried out and data such as vehicle speed,master cylinder piston displacement,master cylinder piston speed and master cylinder pressure are collected.Second,a BP neural network is established using the master cylinder piston displacement and master cylinder piston speed as feature inputs and the real master cylinder pressure as target output.Third,the BP neural network is trained by the training dataset and gradient-descent algorithm.Finally,the pressure estimation performance is verified by the testing dataset.The results show that the proposed algorithm reduces the estimation error by 38%and 15%,compared with the dynamic pressure-displacement model and the LSTM-based estimation algorithm,respectively.

关 键 词：电子液压制动 主缸液压力估计 位移压力模型 BP 神经网络 

分 类 号：U461[机械工程—车辆工程]