基于模糊神经网络的电梯群控系统的研究

发布时间：2018-08-02 13:08

【摘要】：随着社会经济的发展,高层建筑日益增多,电梯群在高层建筑以及智能大厦中所起的作用越来越大,电梯群控系统已成为国内外研究的热点。本文对电梯群控系统的研究主要包括两方面内容：电梯交通模式识别和调度算法的研究,并且在研究中引入了智能控制方法。 首先,本文阐述了论文的课题背景以及研究的目的和意义,回顾了电梯群控的发展与研究现状。 其次,本文研究了电梯群控的基本特性,主要有不确定性、扰动性、非线性和多目标性,并且给出了交通流的基本概念以及检测交通流的方法。研究了电梯群控系统的性能评价指标,主要包括时间评价指标和能耗评价指标。研究了电梯群控系统的构成。 然后,本文研究了应用于电梯群控系统的Mamdani型模糊神经网络,模糊神经网络融合了模糊逻辑和人工神经网络的优点,易于表达知识并且有自学习能力。文中给出的Mamdani型模糊神经网络为交通模式识别与优化派梯提供了理论基础。 根据给出的模糊神经网络对电梯交通流进行模式识别。本文研究了六种典型的交通模式,详述了各个交通模式的特征。采用三阶段混合学习算法对模糊神经网络进行学习,并结合实际交通特点采用两个模糊神经网络对交通流分两步进行识别,先用网络Ⅰ识别出上行高峰、下行高峰、空闲交通以及层间交通的比例,若层间比例较小时不需要进行网络Ⅱ的模式识别,若层间交通比例较大时,运用网络Ⅱ识别出两路、四路以及随机层间交通模式的比例。用样本训练模糊神经网络,并用实际的交通流对模糊神经网络进行测试。 最后,研究了电梯群控调度算法,电梯调度是一个典型的多目标规划问题。本文采用前文提出的Mamdani型模糊神经网络对电梯群进行优化控制,控制目标选择为平均候梯时间、平均乘梯时间、能耗。根据专家规则确定了进行优化派梯的模糊神经网络,采用误差反向传播算法对网络进行学习。通过对实际呼梯信号的调度,进一步验证了算法的有效性。
[Abstract]:With the development of social economy and the increasing number of high-rise buildings, elevator group plays a more and more important role in high-rise buildings and intelligent buildings. The elevator group control system has become a hot spot at home and abroad. In this paper, the research of elevator group control system mainly includes two aspects: elevator traffic pattern recognition and scheduling algorithm, and the intelligent control method is introduced in the research. First of all, this paper describes the background of the thesis, the purpose and significance of the research, and reviews the development and research status of elevator group control. Secondly, this paper studies the basic characteristics of elevator group control, including uncertainty, disturbance, nonlinearity and multi-objective, and gives the basic concept of traffic flow and the method of detecting traffic flow. The performance evaluation index of elevator group control system is studied, including time evaluation index and energy consumption evaluation index. The structure of elevator group control system is studied. Then, this paper studies the Mamdani fuzzy neural network used in elevator group control system. Fuzzy neural network combines the advantages of fuzzy logic and artificial neural network, and it is easy to express knowledge and has the ability of self-learning. The Mamdani fuzzy neural network provided in this paper provides a theoretical basis for traffic pattern recognition and optimization of ladders. According to the given fuzzy neural network, the elevator traffic flow pattern recognition is carried out. In this paper, six typical traffic modes are studied, and the characteristics of each traffic mode are described in detail. The three-stage hybrid learning algorithm is used to study the fuzzy neural network, and two fuzzy neural networks are used to identify the traffic flow in two steps according to the actual traffic characteristics. First, the uplink peak and the downlink peak are identified by network 鈪，

本文编号：2159535