面向产品设计的资源主动服务与优化配置管理研究

发布时间：2018-07-21 11:18

【摘要】：随着企业间的竞争日趋激烈,实现快速、有效、个人化的产品供应是企业发展的必然要求。在产品设计过程中,为缩短产品研发周期和提高资源服务效率,实现设计资源的主动服务,解决设计资源服务需求的主动获取问题,并设法从海量的网络设计资源中找到适合设计人员真正需求的资源进行调度,快速高效地将合适的设计资源主动推送给所需的设计人员。为避免超大规模且复杂多变的数据源给网络设计资源管理系统的维护人员带来的巨大挑战,提出网络设计资源的优化配置与自动化管理方法。针对上述问题,本文做出如下研究:(1)设计资源服务相关的概念与理论基础。对相关的基础理论进行定义及分类分析,为研究如何进行高效的设计资源服务提供了前提条件。深入地了解相关领域中的资源流向特点和传递过程,分别对设计资源、设计资源服务需求和云设计资源三个领域进行定义及分类分析。(2)基于情境感知的设计资源服务需求的主动获取方法。为实现设计资源服务需求的主动获取,在分析情境感知推理层特点的基础上,构建主动获取设计资源服务需求的情境感知服务体系。采用贝叶斯方法使设计资源类别偏好情境化,并根据不同的情境特点,为选择合适的且能融合到推荐中的方法,提出融合资源类别偏好的协同过滤获取算法,通过多个设计资源服务需求的期望值计算及其大小比较,实现了基于情境感知的设计资源服务需求的主动获取。(3)面向设计人员需求与偏好的云设计资源节点的主动反馈方法。提出面向产品设计的云设计资源建模方法,在云设计资源响应能力模型和设计人员需求与偏好模型的基础上,构建基于负反馈的云设计资源调度机制,通过对该调度机制的求解,实现了云设计资源响应能力与设计人员需求的相似度匹配,并将匹配度较高的云设计资源节点反馈给设计人员,提高了资源利用率。(4)云设计资源的自适应优化配置管理方法。建立一种基于神经网络和多目标遗传算法的云设计资源自适应配置模型,利用神经网络预测算法对资源负载进行预测,并根据预测值提出虚拟机迁移请求。为提供最优的虚拟机迁移策略,将基于混合分组编码的多目标优化遗传算法引入虚拟机资源管理,节省了虚拟机迁移时间并减少了物理节点数量,实现了云设计资源的自适应优化配置管理。仿真及结果分析表明,该研究方法能在保证能耗与服务等级协议超标率较低的前提下,提高云设计资源服务效率和质量。
[Abstract]:With the increasingly fierce competition among enterprises, the realization of rapid, effective, personalized product supply is an inevitable requirement for the development of enterprises. In the process of product design, in order to shorten the period of product research and development and improve the efficiency of resource service, to realize the active service of design resource, and to solve the problem of active acquisition of design resource service demand. From the massive network design resources, we try to find the resources that are suitable for the designers' real needs for scheduling, and quickly and efficiently push the appropriate design resources to the designers who need them. In order to avoid the huge challenge to the maintainers of the network design resource management system caused by the large and complex data sources, the optimal configuration and automatic management method of the network design resources are proposed. To solve the above problems, this paper makes the following research: (1) the concept and theoretical basis of designing resource services. The related basic theories are defined and classified, which provides a prerequisite for the research on how to design resource services efficiently. Deeply understand the characteristics and transfer process of resource flow in related fields. Design resource service requirement and cloud design resource are defined and classified. (2) Context-aware design resource service requirement acquisition method. In order to realize the active acquisition of design resource service requirements, a situation-aware service system is constructed on the basis of analyzing the characteristics of context-aware reasoning layer. The Bayesian method is used to situate the design resource category preference. In order to select the appropriate method which can be fused into the recommendation, a collaborative filtering algorithm is proposed according to the different situation characteristics. By calculating the expected value of multiple design resource service requirements and comparing their sizes, we realize the active acquisition of design resource service requirements based on situational awareness. (3) an active feedback method for cloud design resource nodes based on designers' needs and preferences. A method of cloud design resource modeling for product design is proposed. Based on the response ability model of cloud design resources and the demand and preference model of designers, a negative feedback based scheduling mechanism for cloud design resources is constructed. By solving the scheduling mechanism, the similarity between the response ability of cloud design resources and the designer's requirements is realized, and the node of cloud design resources with high matching degree is fed back to the designer. (4) an adaptive optimal configuration management method for cloud design resources. An adaptive resource allocation model for cloud design based on neural network and multi-objective genetic algorithm is established. The neural network prediction algorithm is used to predict the resource load, and a virtual machine migration request is proposed based on the prediction value. In order to provide the optimal migration strategy of virtual machine, the multi-objective optimization genetic algorithm based on hybrid block coding is introduced into virtual machine resource management, which saves the migration time of virtual machine and reduces the number of physical nodes. The adaptive optimal configuration management of cloud design resources is realized. Simulation and result analysis show that the proposed method can improve the efficiency and quality of cloud design resources on the premise of low energy consumption and low rate of service level agreement.
【学位授予单位】：南昌航空大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F272;TB472

【参考文献】