基于遗传算法的模糊神经网络电梯群控-控制理论与控制工程专业论文.docxVIP

摘 摘 要 本文首先对数据融合进行研究。介绍了基于模糊联结聚合神经网络的一种多 信息层次混合结构的融合模型，对网络中每一个节点选择合适的数据融合运算。 建立了进行数据融合的模糊联结聚合网络。 研究了模糊神经网络，用神经网络的结构进行模糊推理。研究了遗传算法及 其实现。采用小生境共享函数法，维护了群体的多样性。用遗传算法优化具有全 局性的模糊神经网络初始参数，用BP算法调节和优化具有局部性的参数。建立 基于遗传算法的模糊神经网络。 在电梯群控系统中，每部电梯采用一个模糊联结聚合网络进行数据融合。采 用二层融合结构：原始数据层融合、特征层融合。对原始层数据采用模糊联结“或” 和“并”运算，特征层采用模糊联结“广义均值”融合运算，融合结果作为基于 遗传算法的模糊神经网络的输入。根据各种交通模式相应电梯群控目标产生模糊 神经网络的样本，训练好的网络用于电梯调度。选取网络进行电梯群控。 在虚拟环境下进行仿真，结果证明基于遗传算法的模糊神经网络电梯群控可 以适用于各种交通流模式，又能满足各单一交通模式下的性能要求。 关键词：电梯群控基于遗传算法酾模糊神经网络数据融合模糊联结聚合网络 ／ 变通流模式 AbstractThis Abstract This paper discussed Data Fusion firstly．A kind of multi—information mixed strueture fusion model based on fuzzy coupl ing aggregation neural networks was introduced whose each node has its appropriate data fusion operator．In this way，a fuzzy coupl ing aggregation neural network which i s used to data fusion was set up． Fuzzy neural network was studied and fuzzy reasoning was real lzed by use of neural networks structure．In addition，this paper studied Genetic Algorithm and its Niche function—shared method which can be used to keep colony diversity．With this understanding。GFNN，genetic—based fuzzy neural network，was made using Genetic Algorithm to ini tial ize general parameter and using BP algorithm to optimize local parameter． In elevator group control system，data fusion according to fuzzy coupling aggregation neural network was executed for each elevator where bi-hiberarchy was adopted—using fuzzy coupling operator OR and operator AND to original data fusion and using fuzzy coupl tag operator General ized—Meaned to eigenvector fusion，and，fusion results were used Hs input of GFNN．Whi le GFNN．whose sample gained according to elevator group control objective of corresponding traffic mode，was trained for elevator di spatching．Therewith，any net trained successf’u11 Y can be chosen to e1 evator group contr01． At 1ast．simulation was made under virtual environment．Elevator group control usi ng genetic—based fuzzy neural network prove to not ONl Y adapted to V