液压挖掘机作业循环阶段的智能识别  

Intelligent identification for working-cycle stages of hydraulic excavator

作　　者：冯培恩[1] 彭贝[1] 高宇[1] 邱清盈[1] 

机构地区：[1]浙江大学机械设计研究所

出　　处：《浙江大学学报（工学版）》2016年第2期209-217,共9页Journal of Zhejiang University：Engineering Science

基　　金：国家"十一五"科技支撑计划资助项目(2013BAF07B04-03)

摘　　要：为了实现对液压挖掘机作业循环各阶段的自动识别,提出以主泵压力为识别对象的智能识别方法.以各阶段开始或结束时的一段波形作为分段标志,对作业循环进行分段.采用有向无环图支持向量机(DAGSVM)多分类方法,识别各分段标志,根据各类样本之间的可分度,优化DAGSVM结构,同时设定距离阈值,保证被识别出的波形为分段标志.引入智能校验系统,对由操作手误动作等引起的识别错误进行校正,使识别准确率由65%提高至95%.最后分析了分段标志宽度对识别准确率的影响.实际测试表明,该方法识别准确率高,实时性好,能够有效识别作业循环各阶段.An intelligent identification method using the pump pressure as the identifying object was proposed in order to automatically identifythe each stage of a working-cycle of hydraulic excavator.The segmentation of a working-cycle was achieved by choosing the pressure curves at the beginning or the end of each stage as the segmentation marks which were recognized by the directed acyclic graph support vector machine（DAGSVM）.The structure of DAGSVM was optimized according to the divisibility between the samples of each class.A distance threshold was set to ensure that the identified curves were the segmentation marks.An intelligent verifying system was introduced to correct the identification errors caused by incorrect operation so that the recognition accuracy rose from 65% to 95%.The relationship between the width of segmentation marks and the recognition accuracy was also investigated.Experimental results show that the proposed method can effectively identify the working-cyclestages,with high recognition accuracy and good real-time performance.

关 键 词：作业循环 识别 主泵压力 有向无环图支持向量机(DAGSVM) 智能校验系统 

分 类 号：TU621[建筑科学—建筑技术科学]