Abstract: Aiming at the problem of pattern recognition of traffic mode in elevator group control system, this paper proposes a pattern recognition method of elevator traffic mode based on multi-value classification Support Vector Machine(SVM). It analyzes the collected elevator traffic flow data using the direct multi-value classification SVM. The traffic mode classifier is established and it can provide an efficient solution for the recognition of non-linear system with multiple-input and multiple-output. Experimental results show that the method can result in global optimization, small classification errors and the ability to meet group control systems.

Key words: multi-value classification, elevator traffic flow, Support Vector Machine(SVM), elevator group control system

摘要： 针对电梯群控系统中的交通模式识别问题，提出一种基于多值分类支持向量机(SVM)的电梯交通模式识别方法。采用直接多值分类SVM对采集的电梯交通流数据进行分析，得到交通模式分类器，从而解决电梯交通流模式识别中多输入、多输出的非线性系统辨识问题。实验结果表明，该方法可实现全局最优且分类误差较小，能满足群控系统的要求。

关键词: 多值分类, 电梯交通流, 支持向量机, 电梯群控系统

CLC Number: 

• TP391