人工神经网络应用于厅堂声级预测  

Neutral Networks Predictions of Speech Levels in Classrooms and Halls

作　　者：罗小华[1] 宁玲 

机构地区：[1]苏州科技学院建筑系,江苏苏州215011

出　　处：《苏州科技学院学报（工程技术版）》2006年第3期73-76,共4页Journal of Suzhou University of Science and Technology （Engineering and Technology）

摘　　要：根据目前已有文献的研究结论和厅堂实测数据,将人工神经网络应用于厅堂声级的预测。结果表明:对于矩形平面厅堂,选择少数与厅堂声级相关性高的几何、物理参量作为神经网络模型的输入变量,可以准确地预测厅堂声级。Using the results of acoustic measure, neutral network analysis is used to predict the speech levels for classrooms and halls. It shows that the good agreements between measured and calculated results can be obtained if the basic parameters used as inputs to the first layer of the neutral network are reasonable.

关 键 词：人工神经网络 厅堂 声级 

分 类 号：TU112[建筑科学—建筑理论]