基于EMD的滚珠丝杠副故障特征提取技术与方法研究

Study on Ballscrew Fault Characteristics Extraction Technology Based on EMD

基于EMD分解得到多个IMF分量,取前10个IMF分量计算能量,通过归一化处理,得到滚珠丝杠振动故障信号能量的归一化特征向量,建立丝杠故障信号各频带能量与丝杠故障之间的映射关系。该向量可作为神经网络输入向量,为神经网络提供了新的输入样本。

Based on EMD decomposition,a number of IMF components were gotten.The top 10 IMF components were taken to calculate the energy.Then through naturalization processing,the ball screw vibration fault signal energy normalized eigenvectors were gotten,the mapping between the screw fault signal band energy and the screw failure was established.The vector can be taken as neural network input vector.It provides new input samples for neural network.