基于FIS、BPNN方法的政府投资项目投资估算方法  

Approach to Estimating the Cost of Government-invested Projects Based on FIS and BPNN

作　　者：余建星[1] 张小平[1] 段晓晨[2] 

机构地区：[1]天津大学建筑工程学院,天津300072 [2]石家庄铁道学院经管分院,河北石家庄050043

出　　处：《工业工程》2007年第6期82-86,共5页Industrial Engineering Journal

基　　金：国家自然科学基金资助项目(70373032);"985"工程资助项目

摘　　要：针对没有类似工程的新建工程项目,提出了投资估算的新方法。将新建工程项目分解为n项单位工程,然后利用反向传播神经网络从大量已完工程历史数据中"提取"类似单位工程,从非线性角度实现对具有类似项目的单位工程造价的预测;对没有类似项目的单位工程,分析其工程特征,将其区分为已知工程特征和未知工程特征,并利用已知工程特征和未知工程特征之间的经验或逻辑关系,建立模糊推理系统,使未知工程特征变为已知工程特征。根据工程特征与工程造价之间的经验或逻辑关系,建立模糊推理系统,计算得到没有类似项目的单位工程造价。A new approach is proposed in order to solve the problem that no existing cost estimation methods are suitable for all new projects. In this paper, a new project is discomposed into ＂n＂ Items and is applied with BPNN（Back-propagation Neural Network）to distill ＂ the similar items＂ from the historical data so that the items of the new project, based on the nonlinear theory, can be estimated. The characteristics of the new project items are analyzed and divided into known and unknown characteristics and, based on their relationship, a fuzzy inference system is built to make the unknown ones known. According to the relationship between the characteristics and the cost, a fuzzy inference system is established to compute the cost of new items. The historical data and the experts＇ experience are fully used in all the methods above in order to distill the similarities, create reasonable rules and provide new methods for estimating the primary cost of Hi-tech projects.

关 键 词：模糊推理系统 模糊逻辑 造价估算 反向传播神经网络 

分 类 号：F830[经济管理—金融学]