基于卷积神经网络的数控机床滚珠丝杠副故障诊断研究

• 作 者: 唐旭 谭继文 徐卫晓 井陆阳
• 机 构: 青岛理工大学,山东 青岛,266520
• 出 处: 煤矿机械
• 发表日期: 2019年1期
• 关键词: convolutional neural network ball screw pair multi-source information fault diag

使用卷积神经网络对数控机床滚珠丝杠副故障信号进行识别.使用多种传感器采集滚珠丝杠副不同种类故障信号, 提取故障信号的特征值作为神经网络的输入.比较卷积神经网络对单一传感器信号特征与多传感器信号特征的识别效果并引入RBF神经网络与BP神经网络作为对比.实验结果表明, 卷积神经网络在多源信息的处理上具有更高的准确率.“,”The Convolutional Neural Network is used to identify the ball screw fault pair signal of CNC machine. A plurality of sensors are used to collect signals of different fault types of the ball screw pair, and the characteristic values of the fault signals are extracted as inputs of the neural network.Comparison of the recognition effects of convolutional neural networks on single sensor signal characteristics and multi-sensor signal characteristics and the introduction of RBF neural network and BP neural network as a comparison. The experimental results show that the Convolutional Neural Network has higher accuracy for the processing of multi-source information.