【作者】 王英；

【导师】 黄磊；

【作者基本信息】 北京交通大学 ， 管理科学与工程， 2014， 博士

【摘要】 摘要：本文以港口物流流程为研究对象,将流程知识挖掘相关的前沿理论技术和智能物流发展中的实际管理问题紧密结合,融合智能物流、港口物流、业务流程管理、工作流建模、过程挖掘和数据挖掘等多种理论和方法,建立了包括港口物流流程模型构建、港口物流控制流分析、港口物流流程诊断分析等三个部分组成的港口物流流程知识挖掘理论方法体系。在此基础上,将物联网智能感知技术、流程知识挖掘和物流模拟仿真技术集成,提出了港口物流流程智能优化设计的集成方法框架。首先,论文通过对智能物流、知识管理、数据挖掘、业务流程管理和过程挖掘等相关理论方法研究现状的梳理,提出由物流控制流知识、物流数据流知识、物流组织知识和物流风险知识四部分构成的港口物流流程知识概念框架,指出物流智能的目的在于应用各种技术降低物流中的各种不确定因素影响及其所带来的风险,分析了现有物流流程分析方法在智能物流发展过程中所存在的局限,明确了物流流程知识挖掘在实现智能物流中的重要支撑作用,从流程知识角度为智能物流的研究提出了一种新的方法思路。其次,论文基于港口物流流程的概念描述和特征分析,指出港口物流流程可以依据组织结构划分为松散结构和高度结构化的两部分。通过分析声明式和程序式建模方法的原理及其适用场合,指出单一的程序式建模方法在对灵活性要求较高的管理业务流程建模时存在局限,提出了融合声明式建模方法和程序式方法的港口物流流程集成建模方法。第三,论文基于模糊挖掘方法,提出了港口物流主干流程探查的方法,并融合港口物流属性信息提出了复杂流程事件日志的日志分组策略。在此基础上,综合运用模糊挖掘和启发式挖掘等各种控制流挖掘技术,提出了包括港口物流事件日志抽取、日志预处理、主干流程探查和子流程划分、事件日志流程实例分组和日志子集控制流挖掘等五个主要步骤的港口物流控制流挖掘分析方法框架。所提出的方法有效地改善了控制流挖掘结果模型的精确度,降低了模型复杂度,提高了挖掘结果的可理解程度,为港口物流流程行为分析提供了有效的智能方法支持。第四,论文研究了港口物流流程中的数据流知识和风险知识挖掘方法,以此为基础提出了基于过程挖掘技术的港口物流流程绩效分析和风险诊断方法。通过改进流程实例聚类算法,有效地改进了无向导学习的过程挖掘结果,实现港口物流流程实例按照流程行为的有效分组,并生成流程实例概貌描述数据集。在此基础上,采用数据挖掘方法挖掘出港口物流属性、流程行为和流程绩效之间的关系模式知识,实现港口物流流程绩效的深度分析。同时,在流程控制流知识和数据流知识挖掘的基础上,将过程挖掘中的一致性检查技术引入物流风险分析中,通过在工作流模型中“重放”事件日志,提出港口物流流程偏差和风险定量分析方法,为实现物流流程智能风险诊断奠定基础。最后,论文基于港口物流流程知识挖掘的研究结果,以广州港集团综合物流管理系统为背景,分析了现有港口物流流程中存在的问题及智能优化流程的必要性。依托物联网等智能技术,讨论了港口物流流程的优化设计及其技术实现方案。通过物联网的智能感知和自动数据采集和传输技术,实现港口物流供应链各环节中货物状态等信息的实时监控和对状态的实时响应及智能应对。最后,将物联网智能感知技术、流程知识挖掘技术和物流仿真技术融合,提出了港口物流流程的智能优化集成方法框架。更多还原

【Abstract】 ABSTRACT:This dissertation sets up a comprehensive framework for the methodology of applying process mining in port logistics in support of the port smart logisitcs. The cutting-edge techniques concering process knowledge discovery are integrated with managerial problems in practice dealing with smart logistics by a combination approach using theories and methods from multiple disciplines including smart logistics, port logistics, business process management, workflow modeling, process mining and data mining. The methodology consists of three main parts including port logistics process modeling, port logistics control flow analysis and port logistics process diagnosis. On this basis, an integrative method for the intelligent design of port logistics processes is proposed combining the IoT sensing techniques, the process knowledge discovery techniques, and the logistics simulation techniques.Firstly, a research review of logistics intelligence, knowledge management, data mining, business process management and process mining is carried out. This makes the basis for proposing the concept framework of the port logistics process knowledge, composing of logistics control flow knowledge, data flow knowledge and logistics risk knowledge. The paper points out that the aim for smart logistics is to reduce the large amount of uncertainties and risks in the logistics processes caused by human-centric activities. The limitations of current logistics process analysis method are then analyzed with respect to the smart logistics development. This highlights the necessity and significance of discovering hidden knowledge in the port logistics processes for support of smart logistics.Secondly, the paper divides the port logistics processes into two parts as loosely-structured and highly-structured, based on the concept and characteristics of port logistics processes analysis. The limitations of imperative workflow modeling for the processes requiring high flexibility are analyzed, and an approach integrating the declarative and the imperative workflow modeling method is thereby presented for the port logistics process modeling.Thirdly, fuzzy mining technique is applied to the process event log to reveal the main flow of the port logistics processes. Using port logistics domain information, the complex event log can be regrouped into several groups. This makes the basis for the detailed control-flow analysis. A comprehensive methodology for the port logistics process control flow analysis is accordingly presented, including event log extraction, pre-processing, main flow exploration and sub process division, instance regrouping, and control flow discovery. A case study is carried out using real data set from an important Chinese port. The result proves the effectiveness of the method in improving the accuracy and reducing the complexity of the model. Consequently, the control-flow analysis is able to provide effective decision support for realizing the smart port logistics.Fourthly, the paper investigates the method for discovering the data flow knowledge and risk knowledge in port logistics processes. The methodology for the port logistics process performance analysis and risk diagnosis is then presented accordingly. By making use of domain information, the trace clustering method is improved and applied for regrouping the cases according to the process behavior. An instance profile generation algorithm is proposed to make the basis for further analysis of the relation between the process behavior and the performance. Data mining techniques are then applied for discovering the knowledge concerning the relationship between the port logistics elements, the process behaviors and the process performance. In addition, a quantitative method is proposed for the port logistics process risk analysis by applying the conformance checking technique. Through’replaying’the event log which records the’real’process behavior in the workflow model which describes the’ideal’behavior, the deviation degree and the activities involved can be revealed. This is a novel approach for risk analysis of port logistics processes.Finally, the problems within the port logistics processes are summarized through the process mining results. The IoT techniques are applied for the improvement of port logistics processes, supporting the real time monitoring of the cargo status information throughout the logistics supply chain. What’s more, an integrative method for the port logistics process improvement is proposed, combining the IoT techniques, the process knowledge discovery techniques, and the data mining techniques.更多还原