Abstract: In industry production processes, there are many serial systems of delay and filter. Joint identification of delay and filter is the cross research area of adaptive system modeling and time delay estimation. For the system model in which the input signal is first delayed and then filtered, this paper develops a pre-delay tracking recursive least squares filtering algorithm based on fast transversal filter realizations. The algorithm is applied to model prediction of a chemical industry process for synthetic ammonia, the simulation results justify that this kind of delay tracking adaptive filter is suitable for low order and raring time delay system modeling.