论文部分内容阅读
提出了一种基于粗糙集属性约简技术的测点优化配置方法。首先根据齿轮箱的故障机理确定了基本测点,采用粗糙集理论建立了测点优化决策表;然后提出了采用基于属性频率的差别矩阵法求取最小属性约简集,避免了复杂的布尔运算;最后通过对约简集进行分析找到了有效的信号监测点,并且应用BP神经网络进行了仿真验证。实验结果表明该方法不需要对监测对象建模,也不需要进行动力学分析,而是根据时频域指标与故障种类之间的关联程度选择有效监测点,通过监控有效监测点,采集有效故障信息,有利于提高故障诊断的效率和准确率。
A method of optimal allocation of measuring points based on rough set attribute reduction is proposed. Firstly, the basic measuring points are determined according to the fault mechanism of the gearbox, and the optimal decision table of measuring points is established by using the rough set theory. Then, the least attribute reduction set is obtained by the differential matrix method based on the attribute frequencies, which avoids the complicated Boolean operations At last, the effective signal monitoring points are found by analyzing the reduction set, and the simulation results are verified by BP neural network. The experimental results show that this method does not need to model the monitoring objects and do not need to do dynamic analysis. Instead, the method selects effective monitoring points according to the degree of correlation between the indicators in time and frequency domains and the types of faults. By monitoring the effective monitoring points and collecting effective faults Information, help to improve the efficiency and accuracy of fault diagnosis.