摘要
路面附着系数是车-路相互作用中最为关键的参数之一,精确识别路面附着系数可用来确定汽车最佳安全控制方式,为此,提出一种基于改进极限学习机(Extreme Learning Machine,ELM)的路面附着系数估计方法。对车辆进行动力学分析,确定神经网络模型的输入量;搭建整车模型和工况,进行仿真试验建立数据集;利用限幅递推平均滤波算法处理数据集,并利用麻雀搜索算法对极限学习机进行改进优化,提高ELM的准确性及稳定性。试验结果表明,改进后的ELM在多方面性能有综合提升,预测准确率为93.4%,提高了4.89%,收敛速度提高了41.33%。
The road adhesion coefficient is one of the most critical parameters in the interaction between vehicle and road.Accurately identifying the road adhesion coefficient can be used to determine the optimal safety control mode for vehicles.Therefore,a method for estimating the road adhesion coefficient based on an improved extreme learning machine(ELM)was proposed.First,the dynamic analysis of the vehicle was conducted to determine the input variables of the neural network model.Then,the complete vehicle model and operating conditions were set up to establish a dataset through simulation experiments.Next,the dataset was processed using a clipping recursive averaging filter algorithm,and the ELM was improved and optimized using the sparrow search algorithm to enhance the accuracy and stability of the ELM.Finally,experimental results showed that the improved ELM had a comprehensive improvement in performance,with a prediction accuracy of 93.4%,an increase of 4.89%,and a 41.33%increase in convergence speed.
作者
康谷峰
张冰战
尹晨晨
边博乾
邱明明
KANG Gufeng;ZHANG Bingzhan;YIN Chenchen;BIAN Boqian;QIU Mingming(School of Automotive and Traffic Engineering,Hefei University of Technology,Hefei 230009,Anhui,China;Anhui Key Laboratory of Digit Design and Manufacture,Hefei University of Technology,Hefei 230009,Anhui,China;School of Mechanical Engineering,Hefei University of Technology,Hefei 230009,Anhui,China;Anhui Research Center for Automotive Technology&Equipment Engineering,Hefei University of Technology,Hefei 230009,Anhui,China)
出处
《农业装备与车辆工程》
2024年第8期33-39,共7页
Agricultural Equipment & Vehicle Engineering
基金
国家自然科学基金项目“自动驾驶接管行为动态迁移规律及接管绩效提升方法”(52172344)
中央高校基本科研业务费专项资金项目(PA2023GDSK0065)
芜湖市科技计划项目(2223jc-04)。
关键词
极限学习机
路面附着系数估计
麻雀搜索算法
神经网络
extreme learning machine
estimation of road adhesion coefficient
sparrow search algorithm
neural network
