摘要
提出一种用时间序列AR模型进行深孔加工状态监测的方法。对所采集的深孔加工过程中的振动信号先进行差分处理,再进行AR建模。模型参数估计使用了方法简单、参数估计无偏、精度高的最小二乘法。然后用模型参数训练一个模式识别神经网络,试验结果表明,这种方法有很好的实用性。
A method that AR model is applied to condition monitoring of deep hole processing is put forward. After difference disposal is applied to vibration signals, AR model is applied to signals collected in course of deep hole processing. The simple, a-gonic and highly accurate least square method is used to estimate the model' s parameters. Then a neural network is trained by the parameters forementioned. The experimental result indicates that the method have good practicability.
出处
《现代制造工程》
CSCD
北大核心
2004年第3期60-62,共3页
Modern Manufacturing Engineering