基于神经网络的RANN竞争模型在内饰视觉工效研究中的应用  

The Application of RANN Competition Model Based on ANN in Visual Performance Research of Cockpit's Interior

作　　者：魏冬冬[1] 陈迎春[2] 孙刚[1] 

机构地区：[1]复旦大学力学与工程科学系,上海200433 [2]上海飞机设计研究所,上海200232

出　　处：《复旦学报（自然科学版）》2014年第5期626-630,635,共6页Journal of Fudan University：Natural Science

基　　金：国家重点基础研究发展计划资助(2010CB734106)

摘　　要：研究驾驶舱内饰材质和光环境对飞行安全的影响具有深刻的意义和实际价值.通过人工神经网络方法分析飞机驾驶舱内装饰视觉工效,其中SOM网络对不同材质进行分类筛选,BP网络对不同属性之间的关系进行拟合和预测.在RANN模型的基础上,应用SOM网络对材质属性进行分类处理,从而提出了RANN竞争模型.RANN竞争模型能够设计出符合舒适度的材质种类,通过对比这两种模型的设计结果,可以发现应用RANN竞争模型对内饰不同属性和舒适度进行研究具有更高的效率,同时验证了此方法具有实际可行性.Research on cockpit＇s interior and ambient light environment has significant meaning and practical value, especially in the field of flight safety. The visual performance research of cockpit＇s interior is based on Artificial Neural Network（ANN）. SOM network is used to classify different material properties, and BP network is used to fit and predict relationship between different material properties. On the basis of RANN model, SOM network is applied to advance a new model which＇s RANN competition model. RANN competition model could be used to design a material variety which accords with visual comfort. Comparing with the results of these two models, it is found that in the research on material properties and visual comfort, RANN competition model has higher efficiency than RANN model. The material varieties which are designed by RANN competition model could accord with visual comfort better. The result shows that using RANN competition model to design has actual feasibility.

关 键 词：驾驶舱 视觉工效 人工神经网络 BP SOM RANN竞争模型 

分 类 号：V250.3[一般工业技术—材料科学与工程] V221[航空宇航科学与技术—航空宇航制造工程]