NJLK机场二期工程软件成本估算研究

发布时间：2018-03-17 10:28

本文选题：成本估算　切入点：神经网络　出处：《南京理工大学》2012年硕士论文　论文类型：学位论文

【摘要】：利用现代信息系统进行项目管理,成本估算起到非常重要的作用,也是项目成本管理的重要基础,一个项目成本管理的成功与否,成本估算起到了决定性的作用。在现有的软件项目成本管理领域中,最主要的问题是软件成本估算方法估算不准确,主观经验影响较大。目前没有一个有效的方法,能够很好的弥补在软件成本估算中的缺陷。本分首先介绍了软件项目成本估算的研究背景和意义,分析了国内软件项目成本管理的研究现状与存在的问题。再介绍了现有的软件项目成本估算方法和软件项目成本管理的概念,并分析了每种方法的特点,总结了现有方法的缺陷。并针对软件项目成本估算不确定性因素多、主观因素影响较大的特点,提出了模糊理论与神经网络方法结合的软件项目成本估算方法,并针对估算结果存在偏差的问题,采用主成分分析法和模糊神经网络与PCA方法结合法对估算结果进行优化,进一步使估算结果更为精确,减少因为训练数据的局限性而引起的结果偏差。通过以上两个步骤的研究工作,建立自己的软件项目成本估算模型,并采用COCOMO模型下的60组典型软件项目训练数据集对估算模型进行多次训练优化,最终形成本文的估算系统。在建立神经网络估算模型的基础上,对NJLK国际机场二期信息工程项目的软件项目开发工作进行了成本估算,包括软件开发工作量估算、软件规模预估、软件开发工作进度估算和软件项目总成本估算。针对多次估算的结果,结合本文提出的主成分分析法、模糊神经网络、PCA结合的方法对估算结果进行了优化。研究表明,该模型能获得较准确和快速的估算软件项目的开发成本,为机场信息工程项目的过程管理提供了重要的依据。
[Abstract]:Using modern information system for project management, cost estimation plays a very important role, is also an important foundation of project cost management, a project cost management success, Cost estimation plays a decisive role. In the current software project cost management field, the most important problem is that the software cost estimation method is inaccurate, subjective experience has a greater impact. It can make up for the defects in software cost estimation. This paper first introduces the research background and significance of software project cost estimation. This paper analyzes the present research situation and existing problems of domestic software project cost management, introduces the existing software project cost estimation methods and the concept of software project cost management, and analyzes the characteristics of each method. This paper summarizes the defects of the existing methods, and puts forward a software project cost estimation method which combines fuzzy theory with neural network method, aiming at the characteristics that there are many uncertain factors and great influence of subjective factors in software project cost estimation. Aiming at the problem of the deviation of the estimation results, the method of combining principal component analysis and fuzzy neural network with PCA method is used to optimize the estimation results, which makes the estimation results more accurate. To reduce the result deviation caused by the limitation of training data. Through the research work of the above two steps, we establish our own software project cost estimation model, The training data set of 60 typical software projects under COCOMO model is used to optimize the estimation model for many times, and finally the estimation system of this paper is formed. Based on the neural network model, the cost of the software project development of the second phase information project of NJLK International Airport is estimated, including the software development workload estimation, the software scale estimation. The software development schedule and the total cost of the software project are estimated. According to the results of multiple estimates and the principal component analysis method proposed in this paper, the fuzzy neural network combined with PCA method is used to optimize the estimation results. The model can estimate the development cost of the software project accurately and quickly, and provides an important basis for the process management of the airport information engineering project.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F562.6;F49;F224

【参考文献】