模糊聚类在数据预处理中的应用研究

Study on Fuzzy Clustering Applied in Data Pre-processing

在分析工业过程数据测量误差源的基础上,提出了一种基于模糊聚类的数据预处理方法。该方法不依赖于任何数学模型,通过对某造纸厂的大量数据的仿真处理,表明该方法能在保留原始数据有用信息的基础上剔除冗余数据,侦破过失误差,减少随机误差。并将该方法用于某造纸厂制浆蒸煮过程的数据预处理中,得到了良好的效果。

Based on the analysis of the sources of the measurement error for process data, the data pre-processing method based on fuzzy clustering is proposed. This method doesn＇t depend on any mathematical model. Through simulation processing on huge amount of data in a paper mill, it is shown that the method keeps valid raw data and eliminates redundant data, detects unpremeditated errors and reduces random errors. The method has been adopted in data pre-processing for pulp cooking process in this paper mil with excellent effects.