电梯群控系统交通需求的迭代学习预测方法  

Prediction of traffic demand for elevator group control system using iterative learning

作　　者：杨祯山[1] 邵诚[1] 马海丰[1] 

机构地区：[1]大连理工大学电子与信息工程学院,辽宁大连116024

出　　处：《控制与决策》2008年第3期302-305,309,共5页Control and Decision

基　　金：国家自然科学基金项目(69874026);国家科技攻关计划项目(2001BA204B01);辽宁省重点实验室基金项目

摘　　要：提出对时变垂直交通需求的智能化迭代学习预测方法.该方法采用最小二乘支持向量机技术,动态预测未来的交通需求及其变化规律.在此基础上,提出以电梯交通需求净增量及其变化率强度为依据的交通模式划分的新规则.通过构造滤波函数的方法获取交通需求变化的主要信息,对交通模式量化识别,得到主要交通模式区间的临界点.仿真实验表明了该方法的有效性.A least squares support vector machine （LS-SVM） based intelligent prediction method is proposed which uses iterative learning to deal with the varying elevator traffic. The future traffic demand is forecasted dynamically to find the varying regular pattern. A new principle of traffic pattern recognition based on the net increment and the intensity of gradient change of the traffic demand is presented. By constructing filter function of the traffic difference, the principal features of the varying traffic demand are extracted and the traffic pattern recognition is conducted to obtain the principal traffic pattern critical points. Simulation experiments show the effectiveness of the proposed method.

关 键 词：电梯群控系统 交通需求预测 迭代学习 最小二乘支持向量机 滤波函数 交通模式识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]