建立未确知RBF神经网络.特点是:综合了未确知系统与神经网络的优点,充分利用已知样本所提供的先验信息,给出了期望输出隶属度的计算方法,网络输出合理且具有良好的可解释性.将未确知RBF神经网络应用于故障诊断领域,取得了很好的效果.
In this paper, unascertained RBF neural network is founded. The features are including the advantages of unascertained system and neural network, using prior knowledge Of the known samples, giving new algorithm to compute membership, and the network output is reasonable and explainable. Using this unascertained RBF neural network in fault diagnosis, the method is effective.