The new mode in multi-mode process can not get enough modeling data in short-term process. And it is hard for
traditional statistical methods to extract process characteristics effectively. Therefore, a new algorithm is proposed to solve
this problem. The initial model is established based on the characteristics of history models. Historical data with the similar
characteristics are found to supply the new modeling data. The iterative model based on the initial model is performed
to fit the process more accurately. A large number of simulations in Tennessee Eastman process show the feasibility and
effectiveness of the proposed method.

limited data modeling;iterative PCA;multi-mode process;process characteristics