Before performing any in-depth analytics, an important issue in dealing with historical dynamic data is to extract data segments from the operation data. The raw data always involve different kinds of patterns over a long series of sequences because of the operation changes or some faults. Conventional methods for identifying the patterns are based on variance or covariance, but the indices designed from variance or covariance are not necessarily able to cover the process dynamics completely, so the precision degradation of segmentations will be induced. This paper designs a new dissimilarity index based on the auto-correlation function to deal with dynamic data. And a fast and high-efficient segmentation algorithm using interval halving and the moving window is proposed to locate the change points of data patterns. Moreover, a new dynamic performance assessment index is designed to find the optimal segment in the dataset. Two case studies, including a numerical example and a distillation column benchmark, validate the advantages of the proposed scheme over the conventional variance-based methods. (C) 2019 Elsevier Ltd. All rights reserved.