摘要
机械设备状态监测与故障诊断技术是保证机械设备安全稳定运行的一项重要措施。由于机械设备结构越来越大型化和复杂化 ,依靠人自身的经验和能力难以判断其征兆与故障之间的关系。随着计算机技术在各个领域的应用 ,智能型的自动监测与诊断技术在机械设备中得到了广泛应用。根据几种典型神经网络特点 ,选择了CP神经网络作为机械故障模式识别器。以大型机组典型故障的频率域特征参数作为网络的训练样本 ,对CP网络进行了训练 ,再将实际的一组频率域特征参数输入到模式识别器中 ,对故障类型进行识别。结果表明 ,以CP神经网络构筑的故障模式识别器有很强的非线性映射能力 ,可对机械设备故障模式进行正确分类。
Mechanical device state monitoring and fault diagnosis technology is an important measure to guarantee mechanical devices running in good state. Because machines' structure become larger and more complex, simply depending on engineers' experience and ability, the relationship between symptoms and faults can not correctly be judged. As computer is widely used, intelligent automatic monitoring and diagnosis technique become popular in machines. Characteristics of several typical neural networks were introduced, and the CP neural network was taken as the recognition system of mechanical fault. Frequency domain characteristic parameters of large machine unit typical faults was taken as training sample to train the CP network, then a set of actual parameters was input to the recognition system to identify the type of the faults. The result indicates that based on CP neural network, the fault pattern recognition system has strong nonlinear mapping ability, therefore it can be used to correctly classify the mechanical faults.
出处
《抚顺石油学院学报》
2002年第4期45-48,共4页
Journal of Fushun Petroleum Institute
