一种基于谱聚类分析的粒子群聚类算法

本文研究了谱聚类中NJW算法的样本最优划分问题.利用粒子群算法在聚类问题上搜索到的全局最优,获得了NJW算法对聚类样本的最优划分.推广了谱聚类算法在样本划分时的普适性和稳定性.实验对比验证该算法是有效的.     

分类属性数据的泛化中心聚类算法

针对采用经典划分思想的聚类算法以一个点来代表类的局限,提出一种基于泛化中心的分类属性数据聚类算法。该算法通过定义包含多个点的泛化中心来代表类,能够体现出类的数据分布特征,并进一步提出泛化中心距离及类间距离度量的新方法,给出泛化中心的确定方法及基于泛化中心进行对象到类分配的聚类策略,一般只需一次划分迭代就能得到最终聚类结果。将泛化中心算法应用到四个基准数据集,并与著名的划分聚类算法K-modes及其两种改进算法进行比较,结果表明泛化中心算法聚类正确率更高,迭代次数更少,是有效可行的。     

基于APC和GPS数据的青奥会期间南京公共交通的调度与优化

为了得到青奥会期间南京市合理有效的公交调度方案,本文针对青奥会场馆、运动员村、旅游点等附近的南京公共交通线路,建立模型与算法.首先,通过APC数据与GPS数据的匹配,对客流数据进行站点匹配预处理,根据已有客流量数据,训练小波神经网络,从而对客流分布情况进行预测,然后基于客流预测结果,采用有序聚类法,实现客流高低峰时段的合理划分.其次,详细分析调度问题的关键所在,以时段总发车次数和乘客等待时间两个因素作为目标函数,将时段最大、最小发车间隔和满载率等作为约束条件,提出基于APC和GPS的公交车辆辅助调度模型,通过遗传算法对模型进行求解,得出不同时段的发车间隔和配车次数,并对模型的性能进行评估.以南京市D7路公交运营线路的实际客流数据为例,采用MATLAB软件进行仿真实验,得出优化结果.结果表明所建模型是合理的,从而为调度时刻表的生成提供了科学的依据.     

我国金融市场日内状态特征聚类与优化交易执行

为探究我国金融市场各日内时段市场状态的规律性与相关性,文章采用一种无监督的聚类方法识别各时段的市场状态特征,并将其应用于优化交易执行问题.文章以中证100全样本股票逐笔成交数据为例,分别在15分钟和5分钟采样频率下构造了反映交易日内不同时段整体市场状态集.通过聚合超顺磁聚类算法,文章实现对训练数据集进行聚类分析,提取市...     

函数型聚类分析:基于距离的一步法框架

基于距离的函数型聚类分析包含曲线拟合和聚类两个独立步骤,最优曲线拟合未必有利于类别信息的提取和保留。根据曲线拟合与聚类分析的计算过程,重新梳理了函数型聚类算法;基于距离度量,提出了同时考虑拟合和聚类效果的函数型聚类一步法;在交替方向乘子法(ADMM)框架下推导并给出了迭代求解算法。模拟试验结果显示,该函数型聚类算法有助于提高聚类精度;针对北京市空气质量监测站点二氧化氮(NO_2)污染物小时浓度数据的实例验证分析表明,该函数型聚类算法对不同类别空气质量监测点具有更好的区分度。     

κ-均值聚类算法的改进及其在冰脊表面形态分析中的应用北大核心

针对传统k-均值聚类算法事先必须获知类别数和难以确定初始聚类中心的缺点,建立了关于聚类中心和类别数k的双层规划模型,结合粒子群算法确定出聚类中心,通过在迭代过程中不断更新准则函数的方法搜索并确定出最佳类别数惫,基于所建模型,提出了一种改进的k-均值聚类算法,并将算法应用于冰脊表面形态分析中.结果表明,算法得到的聚类结果不但具有相邻类别边界清晰的优点,而且能够较好地反映出地理位置和生长环境对冰脊形成的影响.     

κ-均值聚类算法的改进及其在冰脊表面形态分析中的应用

针对传统k-均值聚类算法事先必须获知类别数和难以确定初始聚类中心的缺点,建立了关于聚类中心和类别数k的双层规划模型,结合粒子群算法确定出聚类中心,通过在迭代过程中不断更新准则函数的方法搜索并确定出最佳类别数惫,基于所建模型,提出了一种改进的k-均值聚类算法,并将算法应用于冰脊表面形态分析中.结果表明,算法得到的聚类结果不但具有相邻类别边界清晰的优点,而且能够较好地反映出地理位置和生长环境对冰脊形成的影响.     

遗传模糊聚类算法在图像边缘检测中的应用

将一种改进的遗传模糊c-均值聚类(GFGA)算法应用到图像的边缘检测中.我们将灰度图像中的每一个像素点看成是一个数据样本,将该点的灰度值经过Robert算子、Sobel算子和Prewitt算子处理构成它的特性向量,形成具有三维特征的数据集,然后对这个数据集应用遗传模糊聚类算法进行分类,自适应地检测出图像的边缘点,达到提取边缘的目的.实验结果表明,这种混合算法能得到很好的边缘效果,并且得到的结果无需再细化处理,提高了边缘定位的精度.     

自适应最优混合差异聚类算法

针对多维情况下,数据量很大,不可避免的受到噪声干扰,因此需对原始数据进行必要的处理。本文用二次差异序度调整一次差异度,提出了自适应最优混合差异聚类算法,该算法具有谱系结构,这在地层划分,时间序列分析等应用上极其重要,在地层划分上,本文给出了一个实例,我们可以看到谱系结构的重要性。对无序样品给出了相应的最优聚类算法,并分析了西安地下水文的分布情况。     

基于多目标进化聚类的信用风险特征识别

在信用风险识别领域,聚类算法常被用于区分不同风险等级的样本并识别风险特征。然而该领域中通常面临高维数据处理问题,导致传统聚类算法存在不适应此类问题的缺陷：易陷入局部最优、受冗余特征干扰、鲁棒性不强等。采用高维信用风险数据,研究上市公司信用风险,建立信用风险特征识别的三目标优化模型,设计基于分解的多目标子空间聚类算法进行求解。通过算法的横向对比实验,展示了所提出的算法在聚类精度和鲁棒性方面的优势,并根据聚类算法的权重分配结果,归纳总结上市公司信用风险评估过程中应重点关注的指标。     

Optimizing the release of passenger flow guidance information in urban rail transit network via agent-based simulation

The passenger flow guidance is an effective demand management strategy to alleviate the excessive congestion in the urban rail transit network. In order to determine the scope and the timing, a simulation-based optimization model is proposed to optimize the release of passenger flow guidance information in the rail transit network in this paper. In the optimization model, we mainly focus on three aspects namely; where, when and what type of the guidance information should be released to the passengers. In the simulation model, the passenger choice behavior is captured by the agent-based simulation method, which responses to the congestion and the guidance information. Based on this, the dynamic passenger flow distribution can be derived. Furthermore, the adoption rate of the displayed guidance information on passenger information system as well as its impact on passenger travel behavior are also considered in the model. A hybrid heuristic solution algorithm, integrated with passenger simulator and genetic algorithm, is developed to solve the proposed simulation-based optimization model. Finally, a case study of Beijing subway is carried out with the large-scale smart card data. The numerical study shows that the passenger flow demand affects the guidance effect significantly and the best guidance effect can be met with sufficiently high passenger flow demand. And the guidance rate is also found to affect the guidance results. The results also show that the proposed model can provide a detailed guidance scheme for every station at selected time intervals. The results show that the dynamic releasing scheme can save up to a total of 46,319 min in passenger travel time during a single guidance period.     

城市化进程中地铁晚高峰经济出行模式分析

城市交通有早高峰与晚高峰之分,我国城市晚高峰有区别于早高峰的不同特征。因为晚高峰不需要像早高峰一样考虑直接迟到成本,只需要考虑相对晚到成本,如身体上的疲劳,晚间休息活动时间的减少等因素;并且在大城市,会分流一部分人群在回家前参加晚间活动。本文对晚高峰的这些特点进行分析,通过运筹学最优化方法建立实现社会收益最大化的基于地铁多方式经济出行模型并进行求解,对城市地铁出行经济模式和地铁站联合运营自行车及私家车停车场定价给出了参考性意见。     

Passenger flow control strategies for urban rail transit networks

Urban rail traffic congestion is becoming increasingly serious due to the large traffic demands in modern cities. In order to ensure the safety and quality of station services in peak hours, it's necessary to adopt some reasonable and effective passenger flow control strategies. In this study, through considering the time-dependent passenger demands, a passenger flow control model based on the network-level system is explicitly developed. The passenger successive motion process is discretized by the modeling method. Systematically considering the coordinated relationship between traffic demands and strict capacity constraints (including station passing capacity, platform load capacity and train transport capacity), we establish a mixed integer linear programming model to minimize the total passenger waiting time (including passengers outside stations and on the platforms). The optimization software Cplex is adopted to solve the developed model, and a real network of Beijing urban railway is calibrated to verify the effectiveness of the suggested model. As a result, the proposed flow control strategies can provide detailed information about control stations, control durations and control intensities, and can effectively reduce the total waiting time and relieve the number of stranded passengers in the urban rail transit network.     

周期列车运行图的多目标模型及基于Job-shop的遗传算法研究

计算机编制客运专线周期列车运行图问题已成为国内外研究的热点问题之一.在充分研究国内外周期与非周期列车运行图的规划理论与方法的基础上,构建了我国客运专线周期列车运行图的多目标模型,并将之转化为具有优先级结构的单目标模型;然后结合运行图的数学本质与周期性,设计了基于Job-shop的遗传算法,弥补了国外基于PESP理论所开发的周期列车运行图的算法不足.最后结合京津客运专线实例来验证算法的有效性.     

城市公交查询系统的设计与实现

针对含有“公汽、地铁、步行”的复杂公交网络环境,首先对公交问题所提供的数据进行分析,并优化数据的存储结构;其次充分考虑到公交网络客流分配的主要因素一换乘次数、票价、时间,提出了公交网中这三个目标的加权平均最优路径模型及其算法;最后对模型的算法用Matlab软件实现.通过测试,结果显示本系统能快速响应出满足乘客不同需求的公交出行路径。     

高峰期内的公交服务可靠度研究

在分析高峰期内道路交通现状和高峰客流特点的基础上,从可靠度理论的角度,对乘客在车内的拥挤成本、由于等车产生的时间延误惩罚成本等进行了分析,建立了高峰期内公共交通系统服务可靠度的模型,是评价公交服务能力的一种有效方法.     

A fuzzy clustering approach to real-time demand-responsive bus dispatching control

Quick response (QR) to passenger needs is a key objective for advanced public transportation systems (APTS), and it has become increasingly important for contemporary metropolitan bus operations to gain a competitive advantage over private transportation. This paper presents a real-time control methodology for demand-responsive bus operations that respond quickly to passenger needs. The proposed method primarily involves two levels of functionality: (1) short-term forecasting of passenger demands using time-series prediction models, and (2) identification of service strategies coupled with the associated bus service segments using fuzzy clustering technologies in response to variances in passenger demand attributes and traffic conditions. The proposed bus operations method identifies the demand-responsive vehicle service strategies primarily according to the predicted up-to-date attributes of passengers’ demands, rather than deterministic passenger arrival rates, which were generally used in previous literature. In addition, the variation of traffic conditions along bus lines is considered in the proposed method. Results from numerical studies using real data of passengers’ demands, including passenger volume at each bus stop and the passenger origin-destination (O-D) patterns, are presented to demonstrate the effectiveness of the proposed method for real-world applications.     

Modelling and optimization of average travel time for a metro line by simulation and response surface methodology

This research presents a modelling and solution approach based on discrete-event simulation and response surface methodology for dealing with average passenger travel time optimization problem inherent to the metro planning process. The objective is to find the headways optimizing passenger average travel time with a satisfactory rate of carriage fullness. Due to some physical constraints, traffic safety and legal requirements, vehicle speeds cannot be raised any further to decrease travel time. But travel time can be optimized by arranging headways (i.e. the time period between the departure times of two consecutive transportation vehicles) in a timetable. In the presented approach, simulation metamodels that best fit the data collected from the simulated experiments are constructed to describe the relationship between the responses (average travel time and rate of carriage fullness) and input factors (headways). Then, the Derringer–Suich multi-response optimization procedure is used to determine the optimal settings of the input factors that produce the minimum value of the average travel time by providing a proper rate of carriage fullness. This methodology is applied for a real metro line, and good quality solutions are obtained with reduced number of experiments that needed to provide sufficient information for statistically acceptable results.     

基于实际路网情境的配送车辆调度优化

在实际路网情境下结合车道数、车道宽度、路口信号灯设置等路网物理特性,构建了考虑综合交通阻抗的多车型车辆调度模型,提出了两阶段求解策略:第1阶段设计了改进A-star精确解算法用于计算客户时间距离矩阵;第2阶段针对实际路网的特征设计了混合模拟退火算法求解调度方案。以大连市某配送中心运营实例进行路网情境仿真试验,结果表明:改进A-star算法较改进Dijkstra算法具有更短的路径搜索时间;混合模拟退火算法求解结果较实际调度方案优化了13.1% 的综合成本;路网增流、区域拥堵和路段禁行三类路网情境均能对配送方案的车辆配置、路径选择、客户服务次序、作业时间和违约费用等5方面内容产生干扰,调度计划的制定需要详细考虑这些因素的变化。