摘要
针对神经网络的缺陷和水轮发电机组振动故障原因多、征兆多的特点,利用网络分块技术,把BP网络规模控制在可以接受的范围内,并将专家系统与神经网络相结合,较好地解决了知识获取和自学习的问题。通过实例验证,该网络模型能有效地分离各种故障类型,在水轮发电机组振动故障诊断中具有一定的诊断能力。
Noticing multi-fault cause and multi-sign of turbine-generator unit vibration, and combining expert system with neural network, this paper solve the problem of knowledge acquisition and self-study. To overcome the fault of neural network, partition of network technique is used to control BP network scale in acceptable range. It has been proved via the practical example tests that this system can effectively separate each fault type from each other and that it has a definite diagnostic ability in turbine-generator unit vibration fault-diagnosis.
出处
《西安理工大学学报》
CAS
2003年第4期372-376,共5页
Journal of Xi'an University of Technology
关键词
专家系统
神经网络
水轮发电机组
振动
故障诊断
expert system
neural network
turbine-generator unit
vibration
fault diag-nosis
