论文部分内容阅读
针对人工神经网络训练样本难以获取的困难，提出1种从专家系统获取训练样本的方法，使得产生式人工神经网络能够有效地逼近传统专家系统的诊断推理行为。系统已被应用于125MW汽轮发电机组的故障诊断。
Aiming at the difficulty of obtaining training samples from artificial neural network, this paper proposes a method to obtain training samples from expert system, which enables production-based artificial neural network to effectively approximate the diagnostic inference of traditional expert systems. The system has been applied to the fault diagnosis of 125MW turbogenerator.