基于径向基函数神经网络的工程造价估算

Project investment estimation based on radial basis function neural network

提出了一种更有效的前向网络——径向基函数(RBF)神经网络,以多、高层办公楼为例,建立了工程造价的估算模型,运用MATLAB语言程序实现,同时采用同样的样本对BP网络进行训练,两者结果比较表明,这种方法弥补了BP网络存在的收敛速度慢,易陷入局部最优等缺陷,从而大大提高了其实用性,是对造价估算方式的又一新的尝试。

This paper proposes a new and more effective feed forward network called radial basis function neural network: with high office buildings as examples a project investment estimation model is built. The model is realized by MATLAB. At the same time BP network is trained with the same samples. The two compared results indicate that this method remedy BP network flaw, and improve its application. It is a attempt for the means of the cost estimation and the result is feasible and is reliable.