摘要
为了解决神经网络设计中确定网络结构尤其是隐层单元数的问题,提出了一种基于粗糙集理论确定神经网络结构的启发式算法。通过粗糙集理论属性约简算法对训练样本数据进行处理,根据处理结果确定网络的输入、输出以及隐层单元数。在对A320飞机自动驾驶仪不能衔接的故障诊断实例中,所设计的网络在训练中能够快速收敛,相对于传统的试探法能更快速准确地确定网络结构,从而提高飞机故障诊断效率,缩短维修时间,证明了该方法的实际可行性。
To the problem of designing the neural network structure especially the cryptic layer, a heuristic algorithm is presented based on the rough set theory. The input, output and the cryptic cell number of the network are determined by processing the training sample data of reductiion. An example on the fault diagnosis of A320 autopilot disengagement is provided. The network structure designed according to the presented method can converge faster and more accurate than that by usual tentative method. The simulation result shows the feasibility of the proposed method.
出处
《控制工程》
CSCD
北大核心
2009年第1期42-45,共4页
Control Engineering of China
基金
国家自然科学基金资助项目(60472124)
