【作者】 羊英姿；

【导师】 麦继婷；

【作者基本信息】 西南交通大学 ， 市政工程， 2006， 硕士

【摘要】 随着国民经济的飞速发展，我国的城市化进程不断加快。近年来，各地城市基础设施建设的规模也相应扩大，国家投入了大量的资金进行市政工程建设，从而使得市政工程造价估算与控制受到有关部门及相关人员的关注。工程造价估算是工程造价控制的前提，是工程项目可行性研究的基础，也是招投标制定标底的依据。因此工程造价估算成为工程建设中应首要解决的问题。尽管工程造价估算方法很多，但应用于市政工程方面的估算方法还仅限于传统的估算方法以及时间序列法、因果分析法、回归线性法等，这些方法都存在着一定的缺陷，估算精度不高。鉴于上述，本文主要采用现代数学理论与计算机相结合的三种估算方法即灰色理论法、模糊数学法、RBF神经网络法建立了市政工程造价估算模型。在此基础上，将灰色理论法与神经网络法相结合，提出了灰色神经网络估算模型。针对市政工程项目可行性研究阶段，从定量和定性两方面分析了市政排水工程及市政路面工程的工程特征，找出了影响其工程造价的主要因素，建立工程造价数据库。并用上述四种估算模型对排水工程和路面工程分别进行造价估算分析。分析结果表明神经网络和灰色神经网络估算模型优于其它两种模型。更多还原

【Abstract】 Following the rapid development of the national economy the urbanization process of our country accelerates unceasingly, and the scale of the contraction of urban basic facilities has been enlarged correspondingly. A large amount of fund is thrown into the contraction of municipal engineering, thus the estimation and controlling of the nunicipal engineering cost become the focus of the relavant department and personnel.The engineering cost estimation is the premise of the engineering cost control, also the foundation of feasibility investigation on engineering projects and the basis for tender price. Therefore the engineering cost estimation has turned into a first important problem to be sloved for an engineering construction. Although there are many methods to be used for engineering cost estimation, the methods applied to municipal engineering are limited on some traditional ones, such as the time sequence method, the cause and effect analysis method, the regression method, etc. In all these methods there exist some shortcomings unavoidably, and the low accuracy is the worst of all. For the sake of solving the problem by a new way, the author adopted three non-traditional methods named gray theory method, fuzzy mathematics method and RBF neural network method. On the basis of the work done in order to save computing time a new ’Gray neural network model’ was put forward. In accordance with the work at the feasibility research stage of municipal projects, the engineering characteristics of municipal drainage and pavement projects were analyzed qualitatively and quantitatively. The main factors that affect the engineering cost were affirmed. Comparison among the four models mentioned above shows that the RBF neural network model and the gray neural network model are better than the rest ones.更多还原