基于BP神经网络的建筑工程造价快速预测

快速而可靠的估算工程造价直接影响着招投标决策的合理性。该文根据建筑工程概预算原理,提出以主要特征指标作为典型样本、将字符型转化为数值型特征指标的快速估价方法,在主要特征指标中增加造价指数;并以实例结合带有动量项自适应BP神经网络开展了工程造价的估算。实例计算表明,在考虑造价指数时实例工程造价误差为1.06%,远高于文献[1]的工程造价误差3.02%,提高了工程造价估算精度。该模型具有较高的可行性和可靠性,为工程造价的快速估算提供了一种有效途径。

Fast Estimation for Architectural Project Cost Based on BP Neural Network

The estimation of project cost affects the rationality of project invitation for bid.According to the principles of construction budget estimate,the main character index is identified as a model example,and the char index is transformed into a numeric character index.At the same time, the BP neural network improved by gradient descent momentum and an adaptive learning rate is introduced to estimate the project cost which involved the cost index,and the prototype data is presented and analyzed in the neural network to establish the cost estimation model.The results show that the model is feasible and credible completely, the cost error of which is much better than that in Document [1],and it could provide a more efficient path for fast estimating of the project cost.