摘要
阐述了神经网络模式识别的基本原理,采用改进的BP算法对故障模式识别进行了研究。改进算法采用新的变换函数并引入动量因子,利用变步长算法加速学习,结合汽轮机减速箱故障模式识别进行仿真实验,建立了详细的诊断模型。仿真结果表明,改进算法能够快速收敛,识别结果稳定。
The principle of pattern recognition using neural network was introduced and a new improved BP neural network method for fault pattern recognition was discussed. The improved algorithm uses a new transformation function to expand the range. To accelerate iterative learning, a momentum factor and variable step size method are introduced. The diagnosis model was also established combined with the simulation example of steamer gearbox. The simulation results of fault diagnosis for steamer gearbox have shown that the improved method is fast and effective and reliable identification and can be used for fault diagnosis in other engineering areas.
出处
《机电工程技术》
2008年第10期103-105,共3页
Mechanical & Electrical Engineering Technology
关键词
模式识别
BP神经网络
故障诊断
pattern recognition
BP neural network
fault diagnosis
