群控电梯交通模式识别与调度控制研究

发布时间：2018-03-06 14:24

本文选题：群控电梯系统　切入点：模式识别　出处：《沈阳建筑大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着高层建筑的不断增多,电梯作为现代化建筑中必不可少的垂直交通工具,越来越受到人们的重视,人们对于电梯的性能和服务质量的要求也日益增加。一个合格的电梯系统,在运行的过程中不仅要考虑乘客对楼层去向的需求,还应当考虑到乘客在候梯期间和乘梯期间的心理变化,以及系统能耗等更加全面的问题。计算机技术的发展为智能算法在电梯技术上的应用提供了硬件基础,然而目前没有有效的方法能够合理安排调度电梯资源,电梯系统在这些方面研究还处于起步阶段,尚有很大的研究空间,基于以上目的,电梯的调度算法研究具有重要的社会意义和经济意义。本文首先研究了群控电梯在电梯行业中的地位和发展现状,然后从群控电梯系统出发,通过建立群控电梯系统的数学模型,归纳了群控电梯系统常见的四种交通模式,而后将随机森林算法应用到群控电梯系统中,准确地辨识出了群控电梯系统的交通模式。论文主要包括以下几个方面的内容：(1)分析群控电梯系统工作原理,研究电群控梯系统的发展趋势及亟待解决的问题。通过对各种方法分析比较,最终确定了采用以随机森林算法和匈牙利算法相结合的群控电梯调度方案。(2)以实际电梯为例,通过对在运行过程中客流分布情况进行分析,将群控电梯系统的交通模式分为四种：上行高峰交通模式、下行高峰交通模式、层间交通模式及空闲交通模式。分析这四种交通模式的特点,采用随机森林算法算法对群控电梯进行交通模式识别后,再使用匈牙利算法来确定群控电梯派梯方案。(3)更加深入地分析和研究群控电梯系统,按照不同的交通模式,针对系统的多目标性进行子评价函数的加权,从而构建出具有单目标性质的群控电梯系统数学模型,并同时提出该数学模型的约束条件、决策变量和目标函数。(4)采用随机森林算法对群控电梯系统进行模式识别。通过对数据系统分析,得出电梯系统的交通模式,利用随机森林算法根据数据系统训练一组决策树,将真实电梯数据带入该决策树进行决策分析,由决策树投票决策出当前电梯的交通模式。(5)采用匈牙利算法,对已经由随机森林算法计算出交通模式的群控电梯系统进行派梯方案的设计,以实现对群控电梯系统的多目标优化。(6)通过计算机仿真,验证基于随机森林和匈牙利算法的群控电梯系统模式识别和派梯方案的有效性和优越性。(7)总结本文研究进展,展望下一步工作内容。
[Abstract]:With the increasing of high-rise buildings, elevators, as an indispensable vertical vehicle in modern buildings, have been paid more and more attention to. There is also a growing demand for elevator performance and quality of service. A qualified elevator system should not only take into account passengers' needs for floor movements during operation, Consideration should also be given to the psychological changes of passengers while waiting for and taking the ladder, as well as to more comprehensive problems such as system energy consumption. The development of computer technology provides a hardware basis for the application of intelligent algorithms in elevator technology, However, there is no effective method to arrange the elevator resources reasonably. The elevator system is still in its infancy, and there is still a lot of research space. The study of elevator scheduling algorithm has important social and economic significance. Firstly, this paper studies the status and development status of group control elevator in elevator industry, and then starts from the group control elevator system. By establishing the mathematical model of the group control elevator system, four common traffic modes of the group control elevator system are summarized, and then the stochastic forest algorithm is applied to the group control elevator system. The traffic mode of group control elevator system is identified accurately. This paper mainly includes the following contents: 1) analyzing the working principle of group control elevator system. This paper studies the development trend of the electric group control elevator system and the problems to be solved urgently. Through the analysis and comparison of various methods, it is finally determined to adopt the group control elevator dispatching scheme, which combines the stochastic forest algorithm and the Hungarian algorithm, as an example, taking the actual elevator as an example. By analyzing the distribution of passenger flow in the operation process, the traffic mode of group control elevator system is divided into four types: uplink peak traffic mode, downlink peak traffic mode, downlink peak traffic mode, After analyzing the characteristics of the four traffic modes, the stochastic forest algorithm is used to recognize the traffic patterns of the group control elevators. Then the Hungarian algorithm is used to determine the group control elevator dispatching scheme. (3) the group control elevator system is analyzed and studied more deeply. According to different traffic modes, the sub-evaluation function is weighted according to the multi-objective nature of the system. Thus, the mathematical model of group control elevator system with single objective property is constructed, and the constraint conditions of the mathematical model are put forward at the same time. Decision variable and objective function. 4) the random forest algorithm is used to recognize the pattern of the elevator group control system. By analyzing the data system, the traffic pattern of the elevator system is obtained, and a group of decision trees are trained according to the data system by using the stochastic forest algorithm. Taking the real elevator data into the decision tree for decision analysis, the decision tree votes to decide the current elevator traffic mode. (5) the Hungarian algorithm is used. In order to realize the multi-objective optimization of the group control elevator system, which has been calculated by the stochastic forest algorithm, the elevator system of group control has been designed by computer simulation. Verify the effectiveness and superiority of pattern recognition and ladder allocation scheme based on stochastic forest and Hungarian algorithm) summarize the research progress in this paper and look forward to the next work.
【学位授予单位】：沈阳建筑大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU857

【相似文献】