基于GA-BP神经网络的多传感器轴承故障诊断

由于单一传感器采集滚动轴承的故障信息精度较低,提出基于GA-BP神经网络的多传感器信息融合方法。首先使用单一传感器采集其状态信息,并采用小波包分析提取轴承故障状态特征,然后采用遗传算法(GA)优化BP神经网络对单传感器进行滚动轴承故障诊断,接着运用DS证据理论把每一个诊断结果进行信息融合,最终得到诊断结果。仿真实验结果表明:该方法可提高滚动轴承故障诊断的精确度和效率。

Research on Multi-sensor Bearing Fault Diagnosis Based on GA-BP Neural Network

Considering the low accuracy of making use of single sensor to collect fault information of rolling bearings, a GA-BP neural network-based multi-sensor information fusion method was proposed.Firstly, having the state information collected through a single sensor and the fault state characteristics of the bearing extracted through wavelet packet analysis, and then the genetic algorithm ( GA) used to optimize BP neural network for single-sensor rolling bearing fault diagnosis, finally having information fusion of each diagnosis result imple-mented through DS evidence theory.The simulation result shows that, this method can improve both accuracy and efficiency of rolling bearing fault diagnosis.