摘要
数控机床伺服系统工作原理复杂。为了提高智能故障诊断的准确性,提出基于PSO理论和属性约束规则(RSM)的模糊神经网络算法。先对训练样本进行属性约减;由于PSO算法具有全局优化能力和BP算法具有局部搜索效率高的优点,利用它们训练神经网络,克服了传统方法收敛速度慢的缺点;并通过Matlab仿真证明,该方法具有较高的诊断准确性。
CNC machine tool has a servo system with complex working principles. In order to improve the intelligent diagnose precision, a fuzzy neural network （FNN） algorithm based on particle swarm optimization （PSO） theory and attribute restriction predigesting method （RSM） was proposed for CNC machine fault diagnose. The training samples were predigested with attribute restriction, then the FNN was trained with the method composed of the PSO and BP, for making an exertion of advantages from both the global optimization of PSO and local accurate searching of BP, and the slow convergence shortage of the tradition learning algorithm was overcome. It is proved that the method has better diagnose accuracy through Matlab emulator.
出处
《机床与液压》
北大核心
2012年第1期169-171,共3页
Machine Tool & Hydraulics
关键词
模糊神经网络
粒子群
反向传播
机床故障诊断
Fuzzy neural network
Particle swarm optimization
Back propagation
Machine tool fault diagnose
