摘要
为了在工程项目实施前准确地预测出工期风险的大小,在介绍BP神经网络、遗传算法、主成分分析等理论的基础上,针对现有预测模型的缺点以及BP神经网络自身缺陷,采用主成分分析法对样本数据进行降维处理,并利用遗传算法对BP神经网络的初始权值阈值进行优化,提出了基于PCA-GA-BP的工程项目工期风险预测模型。将以往工程风险数据作为学习样本,训练并构建模型对待建工程项目工期风险进行预测。实例证明该模型有效、可靠,对指导实际工程具有重要意义。
In order to predict the risk of a project time accurately, based on the theories of BP neural network, genetic algorithm and principle components analysis, a model based on PCA-GA-BP to predict the project time risk is established. Considering defects of the existed model and BP neural network, in this model, the dimensions of sample data are reduced by principle components analysis and initial weights and threshold values of BP neural network are optimized by genetic algorithms. The model was trained by historical data and can be used to predict the project time risk. The model is validated by a case
出处
《工程管理学报》
2011年第5期534-538,共5页
Journal of Engineering Management
关键词
工期风险预测
主成分分析
遗传算法
BP神经网络
prediction of time risk
principle components analysis： genetic algorithm
BP neural network
