基于队列挖掘的业务流程时延预测方法与应用

发布时间：2017-12-27 22:32

本文关键词：基于队列挖掘的业务流程时延预测方法与应用　出处：《安徽理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：过程挖掘是业务流程管理的补充,它以事件日志入手提取信息,进而发现、监控、分析和改进实际的业务流程。在技术应用过程中通常结合建模技术使分析过程和结果更加直观。过程挖掘技术在提高企业生产力、指导企业运营和节约运行成本等方面都发挥着极其重要的作用。因此,过程挖掘近年来一直都是研究学者关注的热点课题。随着科技的不断进步,过程挖掘技术也面临着不断的挑战,例如信息系统的不断完善使得业务流程中的数据能够充分的被发现和利用。但是现有的过程挖掘技术大多是基于控制流和数据流角度,针对无时延的业务流程进行基于状态的分析和研究。换句话说,当业务流程中任务的执行存在时延的情况下,现有的过程挖掘技术表现出一定的局限性。针对这一问题,本文提出了基于队列挖掘的业务流程优化分析方法,首先,从事件日志入手,利用行为轮廓的理论结合增量日志挖掘业务流程的初始模型;其次,运用队列挖掘的观点,通过对特定顾客的进行时延预测,挖掘出顾客的行为(时延)信息,以此达到进一步优化过程模型的目的;最后通过实例验证了所提出的优化挖掘方法的有效性。优化后的流程模型不仅对事件日志有很好的重放效果,并且能够反映出多类别且存在时延的业务流程中任务的行为信息。本文主要贡献还包括:(1)提出了业务流程的时延预测多类型队列挖掘方法。在队列挖掘的基础上结合排队论相关知识,通过对G/M/s+M,D/M/c+M和M/M/1三种不同的队列系统进行分析研究,提出了基于这三种不同类别的排队系统的队列挖掘方法。同时将队列系统中不同的顾客类别对时延预测产生的影响考虑在内,对特定顾客进行基于队列长度的时延预测。该方法不仅扩大了队列挖掘技术的应用范围,而且针对目前队列挖掘方法在考虑多队列类型方面的不足进行了完善。(2)提出了基于队列挖掘的服务资源分配方法。服务过程中资源的分配方案是资源如何选取顾客进行服务的问题。合理的资源分配方案在指导服务运行,保证服务质量方面发挥重要的作用。但是目前关于资源分配方案的研究,在考虑资源和顾客交互方面存在一定的不足。本文提出基于队列挖掘的服务资源的分配方法,首先基于领域约束角度,根据资源和顾客的属性进行资源匹配,保证服务过程中资源分配的公平性;然后基于队列角度,考虑流程中资源和顾客的交互,挖掘服务流程中顾客的队列信息和资源的决策信息,对资源分配方案进行优化,提高分配效率。(3)提出了基于配置信息的服务交互过程模型优化方法。服务交互过程模型由服务过程中顾客和资源两方面过程模型交互构成,优化服务交互过程模型对提高资源利用率和顾客的服务体验有良好的促进作用。目前,相关的研究主要关注于资源调度或顾客行为方面,在研究服务交互过程中资源分配对顾客行为的影响方面存在一定的局限性。本文提出了基于配置信息优化服务交互模型的方法,首先提出了服务流程中顾客事件的行为轮廓,从服务流程的服务日志入手,利用行为轮廓的弱序关系分析得到顾客的行为信息;然后基于队列挖掘针对面向服务过程模型结合服务流程中资源的分配方案提出两种新的配置结构,并根据顾客行为信息,利用配置结构对服务交互过程模型进行合理性度分析与行为一致性度检测,通过优化分析获得最终的服务交互过程模型。最后通过实例分析和仿真实验验证了该方法的有效性。
[Abstract]:Process mining is a supplement to business process management. It extracts information with event log, and then finds, monitors, analyzes and improves the actual business process. In the process of technology application, the process and result of analysis are more intuitively combined with modeling technology. Process mining technology plays an extremely important role in improving the productivity of enterprises, guiding the operation of enterprises and saving the cost of operation. Therefore, process mining has been a hot topic of attention of researchers in recent years. With the continuous progress of science and technology, process mining technology is also facing constant challenges. For example, the continuous improvement of information system makes the data in business process can be fully found and utilized. However, most of the existing process mining technologies are based on the control flow and data flow angle, and are based on the state based analysis and research for the business process without delay. In other words, in the case of time delay in the execution of tasks in the business process, the existing process mining techniques show certain limitations. To solve this problem, this paper puts forward the optimization analysis method, business process mining based on queue first, starting from the event log, with the increment of initial log mining model of business process based on behavioral profile theory; secondly, using the queue mining point of view, the delay on specific customer forecast, dig out the customer behavior (in order to achieve delay) information, to further optimize the process model; finally the effectiveness of the proposed optimization of mining method was verified by an example. The optimized process model not only has a good playback effect on event logs, but also reflects the behavior information of multiple categories and time-delay business processes. The main contributions of this paper are as follows: (1) a multi type queue mining method for time delay prediction of business processes is proposed. On the basis of queue mining and related knowledge of queuing theory, we analyze and study three different queue systems of G/M/s+M, D/M/c+M and M/M/1, and propose a queue mining method based on these three kinds of queuing systems. At the same time, the influence of the different customer categories in the queue system on the delay prediction is taken into account, and the queue length based delay prediction is carried out for specific customers. This method not only extends the application scope of queue mining technology, but also improves the shortcomings of queue mining method in considering multiple queue types. (2) a service resource allocation method based on queue mining is proposed. The allocation scheme of resources in the service process is how the resource selects the customer to carry on the service. The rational resource allocation scheme plays an important role in guiding the operation of service and ensuring the quality of service. But at present, the research on resource allocation scheme has some shortcomings in consideration of resource and customer interaction. This allocation method based on queue service resource mining, based on field constraints, resource matching based on the resource and customer attributes, ensure the fairness of resource allocation in the process of service; then based on the queue angle, considering the interaction of resources and customers in the process of mining, customer service process in the queue information and resources decision information on resource allocation scheme optimization, improve the distribution efficiency. (3) a model optimization method of service interaction process based on configuration information is proposed. The service interaction process model is composed of two aspects: the interaction between customers and resources in the service process. The optimization of service interaction process model has a good promoting effect on improving resource utilization and customer service experience. At present, the related research is mainly concerned with resource scheduling or customer behavior. There are some limitations in the research of resource allocation on customer behavior in the process of service interaction. This paper presents a method to optimize the allocation of information service based on interactive model, first proposed the customer behavior profile event service process, starting from the service process of the service log, using weak order relation analysis of customer behavior profile information behavior; then queue mining service oriented process model based on resource allocation scheme in the service process two kinds of new configuration based on the structure, and according to customer behavior information, using the configuration structure of the consistency degree of rationality analysis and behavior detection of service interaction model, service interaction process model by optimizing the final analysis. Finally, the effectiveness of the method is verified by example analysis and simulation experiments.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP301.1;O226

【相似文献】