基于PSO-BP神经网络的掘进机截割部故障诊断

发布时间：2018-12-29 14:19

【摘要】：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。
[Abstract]:In order to improve the validity and accuracy of fault diagnosis of the cutting part of the partial cross-section roadheader, the vibration acceleration data of the cutting part of the roadheader are collected from the coal mine by taking the vibration acceleration signal of the cutting part of the partial cross-section roadheader as the research object. The characteristic vectors representing the running state of the cutting part of the roadheader are analyzed and extracted. The BP neural network is used as the fault diagnosis method, and the weight threshold of the BP network is directly optimized by using the fast convergence and global searching ability of the PSO algorithm. The problem that BP neural network converges slowly and is easy to fall into local minima is solved. By training and testing the data samples, a PSO-BP neural network is constructed to diagnose the fault of the cutting part of the EBZ-160 roadheader, and to diagnose the fault of the cutting part of the EBZ-160 type roadheader. The experimental results show that compared with the BP neural network (FBP) optimized by the fast BP method, the PSO-BP neural network has higher diagnostic accuracy and less training steps. This method can accurately and effectively diagnose the fault of the cutting part of the roadheader, and provides a method and train of thought for the fault diagnosis of the cutting part of the roadheader.
【作者单位】：中国矿业大学(北京)机电与信息工程学院;
【基金】：国家重点基础研究发展计划(973计划)资助项目(2014CB046306)
【分类号】：TD421.5;TP183

【相似文献】