基于神经网络的工程估价系统研究

通过研究工程估价的构成及其影响因素，特别是建筑工程结构和主要分项工程的特征在工程造价中所起的作用，确定了12个主要因素，作为神经网络的输入变量，在深入研究神经网络理论功能的基础上，选用可以应用于工程估价的BP神经网络和BP计算方法，利用S型神经元I／O特性和BP算法的函数逼近功能，使建筑工程造价的估算值的影响因素变化更逼近于其在成本构成中的客观作用，从而更接近于工程实际造价，建立了工程估价模型并开发了计算机辅助系统。收集了60个工程估价实例作为训练样本，10个工程实例作为检测实例例，经测算达到可以用于工程实际的水平。此方法可以快速准确地进行工程估价。

Neural network-based evaluation system for construction projects

Evaluation of construction projects is an important and arduos task for management of construction projects.Ways and means hav e been explored to satisfy the requirements for evaluation of construction projects.With the breakdown of project cost and factors having effect on then,and the functions of structures and their characteristics reviewed in detail,twelve major factors are identified as neural network input variables.The BP neural network and BP calculation method used to evaluate the cost of construction projects are based upon the theory of neural network.the I /O trait of sigmoid nerve cell and the approach function of BP calculation method are used to estimate the cost of construction projects.From the results of theore tical analysis,a model has been established for projcect cost estimation,a lot of actual construction projects are used as training samples and ten projects are used as test samples.Test results show that the precision meets the estimation requirements.Therefore,this method can be easily used to estimate the cost of construction projects.