基于BP神经网络的绿色建筑评价方法及其应用  

Method and Application of Evaluation for Green Building Based on BP Neural Network

作　　者：齐宝库[1] 牛驰野[1] 赵景明[1] 邵帅[1] 

机构地区：[1]沈阳建筑大学管理学院,辽宁沈阳110168

出　　处：《沈阳建筑大学学报（社会科学版）》2012年第2期152-156,共5页Journal of Shenyang Jianzhu University:Social Science

基　　金：住房和城乡建设部科学技术计划项目(2011-R3-27)

摘　　要：将人工神经元网络应用于绿色建筑评价中,建立了采用LM算法的三层BP神经元网络绿色建筑评价模型,并参考国内外对绿色建筑评价的研究,建立绿色建筑评价指标集,通过专家打分获得训练样本,用trainlm训练函数训练网络,验证了绿色建筑评价结果与实际的一致性。最后,通过已训练好的网络,对多个绿色建筑方案进行评价选优,验证了该模型在绿色建筑评价中的实用性与可靠性。The paper applies the artificial neutral to green building evaluation. The model of evaluation for green building based on three-layer BP-network is established by using LM algorithm. By domestic and foreign research for green building evaluation, the evaluation index collection of green building is selected. The network is trained by trainlm function with training samples that have been obtained by expert. The re- sult is consistent with the actual work by the network and the applicability of the evaluation model is illus- trated by examples. Finally, this paper chooses the best one of several green building schemes by using trained network. The practicability and reliability of the model can be verified in this process.

关 键 词：绿色建筑评价 绿色建筑评价指标集 BP神经网络 梯度下降法 阀值 

分 类 号：TU71[建筑科学—建筑技术科学]