高层建筑电梯智能交通流预测理论及关键技术研究

项目名称： 高层建筑电梯智能交通流预测理论及关键技术研究

项目编号： No.60874077

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 万健如

作者单位： 天津大学

项目金额： 30万元

中文摘要： 对城市道路交通问题各国都非常关注，已形成相关理论及控制方法；而对大楼内电梯交通预测问题研究很少，缺乏相关理论和行之有效的控制方法，以至于许多已建大楼电梯配置不合理。电梯配置过高造成浪费；配置过低，乘梯和候梯时间太长，电梯服务质量下降。 针对电梯客流的随机性，各楼层人数分布的不均匀性、楼层高度的不一致性等因素，提出基于粒子群优化径向基神经网络算法；并运用统计学理论中的结构风险最小化原则训练网络；对预测模型进行动态模拟运行、周期性学习和参数修改。 通过在预测模型中设置映射变量，以适应不同高层建筑交通流预测。提出基于多分辨率正交多小波神经网络算法，分别建立五种典型时段交通模式，实现预测模型与五种交通模式的模型嵌套。以平均候梯时间、平均乘梯时间和目的层重复度为控制目标，提出外呼信号智能化群控策略。 研究在风摆、火灾、地震等特殊状况下的运行模式，提出相应控制策略和减灾措施，提高运行安全和舒适感。

中文关键词： 电梯配置；交通流预测；径向基神经网络；正交多小波神经网络；模型嵌套

英文摘要： Countries in the world are all concerned with the problem of the urban road traffic,and it has been formed some related theories and control methods. However,the research on the approach for elevator traffic prediction is less.Now, related theories and effective control methods are so poor that the elevator configuration of many completed building is unreasonable. Too high quality of elevator configuration lead to waste ,on the contary, lead to decrease of service quality. On account of various factors about elevator traffic ,such as the the randomness of the passenger flow, the uneven distribution of number on each floor, inconsistency of floor height etc,a algorithm based on the particle swarm optimization RBF (Radial Basic Function)neural network is proposed. And a principle named structural risk minimization is used to train the net and then a predicted model is used to simulate , to learn and to modify the parameters . By setting up mapping variables in prediction model , this algorithm adapt to the different predictive high-rise building traffic flow. Based on multiresolution orthogonal wavelet neural network algorithm,this paper established five kinds of typical time traffic mode respectively,making predicted model which to be nested in the five types of traffic patterns model come true.With average waitting time, average time of in elevator and the repeat degree of layer as the goal, an elevator with outbound signal Group-Control Policy is proposed. To research on the mode of elevator in special condition ,such as wind, fire or earthquake ,then put forward the corresponding control strategy to improve the operation safety and comfort level.

英文关键词： Elevator configuration；traffic flow prediction；RBF neural network；orthogonal multiwavelet neural network；nested model