Application of data mining methods in eluxyl process

Eluxyl process is a simulated moving bed process. It achieves PX (paraxylene) separation from its other aromatic isomers. One year's historical data of Eluxyl process were studied, and data mining methods were introduced to extract useful operation information from them. Three main data mining procedures were discussed. Data preprocessing was discussed first. Then PCA (Principal Components Analysis) and K-means clustering technology were introduced and applied to discover different operational conditions in different loads. Finally fuzzy classification based on associations (FCBA) algorithm was used to discover the main reasons that lead to performance fluctuations and the rules denoting the typical operations.