LMBP预测模型在国际工程索赔上的应用

Application of LMBP predictive model in international construction claims

针对传统BP算法的缺陷,提出了一种采用L-M算法来加快收敛速度改进的BP神经网络。在此基础上,通过研究工程中引起索赔的各种因素,建立了基于LMBP神经网络的非线性系统,并利用该网络模型来预测工程索赔出现的可能性,并通过具体的仿真以及实践结果验证了LMBP网络的有效性,为承包商的工程索赔管理提出了一个新途径。

Aiming at the shortages of the conventional BP neural network, this paper proposes a new Levenberg- Marquardt（LM） algorithm which speeds up the training of BP neural network. According to these causes of international construction claims, the predictive model of the nonlinear system is set up based on LMBP neural networks, and utilize this network model to forecasts the possibilities appeared in construction claims. By means of simulation and practice, the effectiveness of the LMBP neural network has been further testified, and advances a new approach for management of construction claims.