School of Electromechanic and Information,China University of Mining and Technology ( Bejing)Shijiazhuang Coal Mine Machinery Company Limited

为了提高部分断面掘进机截割部故障诊断的有效性和准确性,提出基于多传感器信息BP神经网络的掘进机截割部故障诊断方法,对EBZ-160型掘进机截割部是否发生故障进行诊断。利用自适应学习速率法和附加动量法相结合的方法(快速BP法),解决BP神经网络原有算法收敛速度慢和存在局部极小值的问题;利用多个传感器采集掘进机截割部状态信号,并通过对掘进机截割部状态信号进行数据分析,提取多组特征向量并建立了掘进机截割部特征数据库。运用BP神经网络对样本数据进行训练,实例分析结果表明,该方法能够有效地监测并诊断掘进机截割部健康状态,诊断精度和准确率较高。

in orderto improve the falure diagnosis eficiency and accuracy of the cuting unit in the roadheader:the paper provided the falure dlagnosis method ofthe cutting unt in the roatheader based on the BP neural network of the mui sensor information and the falure diagnosis was conduced on the cutng unit of the EB.160 nmode roadheader.The method combined with the acaptive leamning rate method and the aditional momentum method was applied to soive the slow convarcence rate of the previous algorithm and the local mininmum problem existed in the BP neural network.The muti sensors were appied ocollect the state sgrals of the cuting .nit in the roacheader.Vith the data analysis conducted on the status signal ofthe cuting unti n the roadheade,muiti roups of the feauwre vectors werecalected an te feature database of the cutting unit in the roadheader was established.The BP neural network was applied to the training of the sample data.The case analysisresut.showed that the method could effectivlly monitor and diagnose the heath status of the cutting unit in the roadheader and the diagnosis precision an accuracy were hgh.

国家重点基础研究发展计划(973计划)资助项目(2014CB046306)；