AT换挡执行机构自学习控制策略研究  

Research on Self-learning Control Method of AT Shift Actuator

作　　者：刘向前 闫娟 杨慧斌 贾茜伟 LIU Xiang-qian;YAN Juan;YANG Hui-bin;JIA Xi-wei(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学机械与汽车工程学院,上海201620

出　　处：《软件导刊》2021年第5期123-127,共5页Software Guide

摘　　要：换挡执行机构在与不同型号AT变速箱安装、匹配时,存在初始档位误差角度值,因此提出基于改进卡尔曼滤波与二分类逻辑回归的算法,设计可匹配不同AT自学习误差控制策略的双芯片架构。首先采用改进的二分类逻辑回归算法,采集处理AT档位数据信号;然后采用改进的卡尔曼滤波算法,有效实现执行机构角度的自行补偿与修正;同时,采用双芯片架构控制模式完成执行机构功能安全性特殊要求;最后,基于CodeWarrior构建测试平台及运行环境,对自学习控制策略进行仿真试验。结果表明,该自学习控制策略可自行补偿修正档位误差值,且可将误差率有效控制在1.87%以内,极大提升换挡机构与AT匹配的普适性与可靠性。To solve the problem of the initial gear angle error value when a current shift actuator is installed and matched with different types of AT gearboxes,this paper proposes an improved Kalman filter and two-class logistic regression algorithm,and designs a twowire architecture which can match the self-learning error control strategy of different ATs.First,to improve classification regression for logistic method was adopted to further realize the acquirement as well as actioning of AT gear signals.Then,improved Kalman image strategy was used to imitate and verify the extraction of a crown-tooth diagram of an actuator;at the similar time,this design was adopted a control mode of a two-chip architecture completes this special requirements of functional safety of an actuator.Finally,a quiz platform and operating environment are built set up CodeWarrior to simulate the designed self-learning control strategy.The results explained that the self-learning control strategy eliminates files,what’s more,which can be controll the initial error angle value within 1.87%,which significantly improves the universality and reliability of the shift mechanism.

关 键 词：换挡执行机构 自学习 二分类逻辑回归 CODEWARRIOR 双芯片架构 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]