【作者】 崔文华；

【导师】 刘晓冰；

【作者基本信息】 大连理工大学 ， 控制理论与控制工程， 2013， 博士

【摘要】 在日益激烈的国际市场竞争环境中,金融机具行业企业面临着尽快提供适合不同层次和群体的产品设计问题。纸币清分机是金融领域现金处理业务的高端必备设备,具有结构复杂、智能化程度高、市场需求变化快的特点。本文针对面向大批量定制的纸币清分机产品族设计过程中一些关键问题进行了深入研究,目的是以最快的速度和最低的成本为客户设计出满足个性化需求的产品,结合纸币清分机开发过程特点,应用产品族设计理论,详细阐述了纸币清分机产品族设计开发过程,并提出了产品族设计过程的配置优化方法,并对研究成果进行了验证和应用。本文的主要研究内容叙述如下：(1)应用基于设计公理的产品族设计理论提出纸币清分机产品族设计过程。建立产品开发过程的多视图模型和任务结构模型,实现任务结构模型到Petri网模型的映射定义并进一步推导与工作流网的映射转换关系,结合纸币清分机的设计开发过程完成了任务结构模型到工作流网的转化并进行了实例验证分析。(2)提出一种面向产品族设计的纸币清分机产品平台核心模块的确定方法。通过建立纸币清分机产品的模块关联矩阵,采用变个体长度的混合蛙跳算法优化模糊聚类数和聚类中心,求得产品平台构成模块的最优模糊划分,并应用切断算子和拼接算子对个体进行重组形成新个体并进行局部寻优,解决模糊C均值聚类局部极小值和初始化选值敏感性问题。在纸币清分机产品族平台设计中进行了仿真验证,实现纸币清分机模块化产品族公共平台和核心模块的有效确定。(3)提出了产品族设计过程中产品族配置性能预测方法。在基于递阶支持向量机的配置综合性能预测方法框架下对纸币清分机产品族的历史数据库进行数据挖掘,预测新配置产品的性能,提出一种改进混合蛙跳算法来优化SVM模型的核函数参数和误差惩罚因子,应用模拟退火算法提高算法的局部搜索能力和收敛速度,通过以纸币清分机产品族为实例的仿真实验并对比其他方法验证所提预测方法的有效性,为准确、快速地进行纸币清分机产品族配置提供了有效方法,满足多样化的客户需求。更多还原

【Abstract】 The financial machinery enterprises are facing with the problem providing proper product design for different kinds of customers especially under the increasingly fierce competition in the international market. The currency sorting machine is important and necessary cash processing equipment in the financial industry. It has the characteristics of complex structure, intelligence and changing demand.In order to meet the personalized products demand as quickly as possible and at the lowest cost, the dissertation carried out an in-depth research on some key problems of the product family design of banknote sorting machine in the process of design for mass customization. The dissertation presents new methods combining banknote sorting machine structure, design features and development processes. The research findings are applied successfully in practice. The main research contents in this dissertation are described as follows:(1) The development process of banknote sorting machine is analyzed in detail taking into account product structure and features based on the product family design-oriented theory and key technology. The multi-views modeling method and the task structures in the product development process are proposed. The mapping mechanism from the task structure of the product development process to the workflow net is set up. The design verification results of banknote sorting machine development process based on WF-NET show that the proposed method can effectively realize the task distribution and performance estimation for analysis the product development cycle accurately and quantitatively.(2) The research on the method for determining product platform design modules in the process of modularized product family design is carried out based on fuzzy C-means (FCM) clustering algorithm optimized by the improved shuffled frog leaping algorithm (SFLA). The component coefficient matrix based on the datum collected is set up. The number of fuzzy clustering and cluster centers are optimized by sizable-individual SFLA to obtain the optimized fuzzy partition of the components. The simulation results of the product family design of the banknote sorting machine show that the proposed method supplies the theory basis of the quantity mathematic analysis and fast configuration for the product family design. (3) A method of product family configuration performance prediction is proposed in the process of product design. The new variant product performances are predicted through the data mining on the historical datum of banknote sorting machine product family under the framework of configuration comprehensive performance prediction method based on SVM. An improved shuffled frog leaping algorithm (SFLA) is adopted to optimize the kernel function parameters and error penalty factors of the SVM models, where the simulated annealing algorithm is used to increase local searching ability and convergence velocity.Finally, the proposed method is verified on a newly developed module-based banknote sorting machine product family to assess the effectiveness of predicting the configuration performances of the module-based product family design accurately and quickly.更多还原