摘要
为提高机械设备的维修诊断能力,提出了基于神经网络的智能诊断技术。采用自适应调整步幅的BP算法,改进后的算法,在误差精度高时能极大提高收敛速度,减少误判率。利用12150L发动机进行了实验,设置典型故障,采集声音信息,对原始声音信号进行解读、采样和分析,形成样本,通过学习和测验,该系统对预设的故障都进行了成功的判断,达到了预期的目的;而且智能诊断系统能分清熟练工人都无法直观判断的故障。实验结果表明,基于神经网络的智能诊断是可行的,具有良好的推广应用价值。
Ability to maintain machine equipments should be improved if the technique of intelligent diagnosis based on neural networks is adopted.An improved BP algorithm which can adjust step parameters adaptively is used to get high speed and reduce the error rate when the er- ror accuracy is high.The audio information is collected by establishing the typical breakdown in 12150L engine.Then original audie signal is disposed to get sample.All of pre-established breakdowns can be recognized successfully.Furthermore,the system can recognize the break- downs which wall-trained worker can't identify directly.So the intelligent diagnosis based on neural networks has good application value.
出处
《控制工程》
CSCD
2006年第S2期201-203,共3页
Control Engineering of China
关键词
神经网络
BP算法
智能诊断
发动机
neural networks
BP algorithm
intelligent diagnosis
engine
