基于决策树优化粗糙集的设备故障诊断指标集约简的方法研究

发布时间：2018-03-30 12:11

本文选题：属性约简　切入点：决策树　出处：《中国煤炭》2016年02期

【摘要】：特征指标集的约简对于机电设备运行状态准确可靠的监测是非常重要的,为此设计了一种基于粗糙集的属性约简方法。该方法首先将时域及频域信号特征作为条件属性,故障类型作为决策属性,建立故障决策表。利用决策树不纯度的减少量对决策表中的连续属性数据进行离散化处理,然后采用粗糙集对离散化后的决策表进行约简,约简后的故障规则决策表可以使设备的故障诊断变得更为简单有效。最后分别采用实验室轴承数据和矿用带式输送机轴承数据设计试验进行验证,试验结果表明,在保证设备状态监测结果可靠的情况下,该方法能够有效地对属性指标集进行约简,降低工作计算量和减少不确定因素的影响。
[Abstract]:The reduction of feature index sets is very important for the accurate and reliable monitoring of the operating state of electromechanical equipment. A rough set based attribute reduction method is designed. Firstly, the time domain and frequency domain signal features are taken as conditional attributes in this method. Fault type is used as decision attribute, fault decision table is established, continuous attribute data in decision table is discretized by reducing the unpurity of decision tree, and then the discrete decision table is reduced by rough set. The reduced fault rule decision table can make the fault diagnosis of the equipment more simple and effective. Finally, the laboratory bearing data and the mine belt conveyor bearing data design test are used to verify, the test results show that, The method can effectively reduce the attribute index set, reduce the work calculation and reduce the influence of uncertain factors on the condition that the equipment condition monitoring results are reliable.
【作者单位】：中国矿业大学(北京)机电与信息工程学院;中国电信江苏分公司;天地(常州)自动化股份有限公司;
【基金】：天地科技科研项目(2014-TDGZZD-01)综采挖掘工作面装备状态监测与故障诊断系统研究
【分类号】：TD40;TD60

【相似文献】