论文部分内容阅读
矿用球磨机故障诊断是典型的复杂工业过程多维数据挖掘问题,难点在于多维数据挖掘准确度低且算法时间复杂度高,为此提出基于局部权重角度离群算法(LW-FastVOA)的数据挖掘方法.首先采用角度离群算法(ABOD)在多维空间中衡量数据点的离群度,并针对ABOD算法时间复杂度算法较高问题,采用FastVOA算法将数据集正交投影于随机超平面上,利用AMS草图推导出各点的方差,归纳将其投影到随机超平面上作为频矩参数,算法的时间复杂度降低.最后提出LWFastVOA算法增加数据点的局部权重,降低多聚簇间离群点遗漏率,从而提高了算法精度.仿真实验结果表明,所提出的LW-FastVOA算法提高了精确率与召回率,验证了算法的有效性和可行性.
Mining ball mill fault diagnosis is a typical multi-dimensional data mining complex industrial process problems, the difficulty lies in the low accuracy of multidimensional data mining and the algorithm time complexity is high, so based on local weighted angle outlier algorithm (LW-FastVOA) data mining method Firstly, the outlier of point data is measured in multidimensional space by using angle outlier algorithm (ABOD). In order to solve the high time complexity problem of ABOD algorithm, FastVOA algorithm is used to orthogonally project the data set onto a random hyperplane. AMS sketch derives the variance of each point, induces it to the random hyperplane as the frequency parameter, and reduces the time complexity of the algorithm.Lastly, the LWFastVOA algorithm is proposed to increase the local weight of the data points and reduce the missing of outliers The accuracy of the algorithm is improved.The simulation results show that the proposed LW-FastVOA algorithm improves the accuracy and the recall rate, and verifies the effectiveness and feasibility of the algorithm.