This paper studies an online identification algorithm for batch processes incorporating priori process knowledge of pole and zero positions. The knowledge is available to control engineers and can be exploited to improve the accuracy of the identified process model. To reduce the computation burden of directly invoking Lyapunov inequality, a bound on the identified parameters is imposed to enforce the match between the priori knowledge and identified model. The bound is recursively calculated according to the newly obtained model. The proposed identification method uses the information not only from the time direction but also along the batch direction to improve the identification performance from batch to batch. A filter is introduced to suppress the variation on the identified parameters. Finally, numerical simulations verify the performance and robustness of the proposed method. (C) 2016 Elsevier Ltd. All rights reserved.