摘要
提出一种基于神经网络集成的水下航行器声学故障识别方法。该方法把经单独训练的具有一定差异度的单个RBF神经网络加以集成,可提高分类器的泛化性能。实验证明该模型的有效性。
A method for acoustic fault identification was presented based on neural network ensemble. In the method ; those single RBF neural networks which had been trained solely were integrated. It improved generalization performance of classifier. The experiment results show the model is effective.
出处
《船海工程》
2009年第2期86-89,共4页
Ship & Ocean Engineering
基金
国家自然科学基金(50775218)
关键词
神经网络集成
小样本
泛化性
差异性
neural network ensemble
small data set
generalization
diversity