基于PSO-BP神经网络的掘进机截割部故障诊断

Fault diagnosis on cutting unit of mine roadheader based on PSO-BP neural network

为提高部分断面掘进机截割部故障诊断的有效性与准确性,以部分断面掘进机截割部振动加速度信号为研究对象,从煤矿井下采集了掘进机截割部振动加速度数据,分析并提取了表征掘进机截割部运行状态的特征向量,采用BP神经网络作为故障诊断方法,利用PSO算法的快速收敛性及全局搜索能力直接对BP网络的权值阈值进行优化,解决了BP神经网络收敛速度慢及易陷入局部极小值的问题。通过对数据样本进行训练与测试,构建了能够诊断截割部是否故障的PSO-BP神经网络,对EBZ-160型掘进机截割部是否发生故障进行诊断。试验结果表明:与快速BP法优化的BP神经网络(FBP神经网络)相比,PSO-BP网络诊断精度更高,训练步数更少。该方法能准确有效地诊断掘进机截割部故障,为掘进机截割部故障诊断研究提供了方法与思路。

In order to improve an efficiency and accuracy of the fault diagnosis for the cutting unit of the mine roadheader,based on the vibration acceleration signal of the cutting unit on the mine roadheader as the study object,the vibration acceleration data of the cutting unit on the mine roadheader were collected from an underground mine and the feature vectors to represent the operation status of the cutting unit on the mine roadheader were analyzed and picked up. The BP neural network was applied as the fault diagnosis method. The rapid convergence and overall discovery capacity of the PSO algorithm was applied directly to the optimization on the weight threshold value of the BP neural network and the slow convergence speed of the BP neural network and the easy falling in the local minimum problem were solved.With the training and test conducted on the data samples,a PSO-BP neural network was established to diagnose the cutting unit whether or not in fault and the diagnosis was conducted on the fault whether or not occurred in the cutting unit of the EBZ-160 roadheader. The test results showed that in comparison with the optimized BP neural network（ FBP neural network） of the rapid BP method,the PSO-BP neural network would have higher diagnosis accuracy and the training steps would be less. The method could accurately and effectively diagnose the fault of the cutting unit on the roadheader and could provide a new method and new idea to the study on the fault diagnosis of the cutting unit on the roadheader.