摘要
电梯交通模式识别是电梯群控调度的一个关键问题.针对实时变化的电梯交通流数据,提出一种新型的电梯系统交通模式识别方法.在不增加数据采集量的基础上,首先对基本交通信息进行预处理,再采用多值分类的支持向量机算法,对电梯群控系统建立交通模式分类器.建立的分类器可以根据交通流数据的变化,自适应地识别出建筑物内的最大客流层及次大客流层(厅堂除外).仿真结果表明,这种交通模式识别方法能较准确地辨识出各种交通流模式,并且通过对比试验,证明该算法的识别准确率优于人工神经网络算法,体现出较好的泛化能力,实用性强.
The traffic pattern recognition is essential in elevator group control systems. Based on the varying elevator traffic flow data, a new elevator traffic flow pattern recognition method is presented. Without additional assignment, original traffic data is pretreated, and then an elevator traffic flow pattern recognition classifier based on multi-value support vector machines is built. The proposed pattern classifier can adaptively recognize the largest traffic flow floor in the building. Experimental results demonswate that this method can recognize different traffic modes accurately and is better than the artificial neural-network method.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2005年第6期900-904,共5页
Control Theory & Applications
基金
国家自然科学基金资助项目(69684001)
关键词
支持向量机
电梯群控系统
交通流
模式识别
support vector machine
elevator group control systems
traffic flow
pattern recognition