改进K-Means算法在油水界面测量中的应用

Application of Improved K-Means Algorithm for Oil-water Interface Measurement

借助物位传感器测量原油储罐内油水界面时,为了获得精确的测量结果,根据油水界面测量数据分布特性,提出了基于K-means算法的油水界面测量计算方法,分析K-means算法在油水界面测量中的基本思想和应用过程。因K-means算法本身无法处理油水界面数据中存在伪数据的问题,又提出了改进K-means的油水界面测量预处理聚类算法,该算法首先建立中值预处理模板,遍历油水界面数据,再对油水界面数据进行聚类划分,确定最优聚类结果,计算油水界面及液位高度。通过实验将该算法应用于存在伪数据的油水界面计算过程中,实验结果表明:改进K-means算法可以有效解决伪数据问题,并能提高油水界面计算结果的准确率,有较小的迭代次数和运行时间,性能优于K-means算法。

In order to achieve the accurate results when measuring the oil water interface by level sensors,a method of oil water interface calculation based on K-means algorithm was proposed according to the distribution characteristics of the data,and the basic idea and application process of K-means algorithm for oil water interface measurement was analyzed.However,because the K-means algorithm can not deal with the false data in the oil water interface data,another new method of preprocessing clus tering algorithm for oil water interface measurement based on K-means algorithm was proposed.The algorithm firstly made a medi an preprocessing template and traversed the oil water interface data,then the algorithm was applied to the calculation of the oil water interface,and finally the oil water interface and liquid level was calculated with the optimal clustering results.Experimental results demonstrated that the improved K-means algorithm is better than K-means algorithm with higher accuracy,smaller itera tion times and running time,and performance is better than K-means algorithm.