To realize smooth and high speed continuous casting, the urgent problem is to control the sticking-type breakout efficiently. In addition to improving casting process conditions of inducing stickers, the key is to develop active and accurate breakout prediction and healing control technology. In this paper, a novel logic judgment model for predicting stickers is presented, which is designed for more rows of thermocouples (TCs) in high density, and uses temperature change rate and vertical and horizontal detections to well recognize sticker propagation. Simulation results indicate that new model can accurately and timely detect all stickers, and has a good robustness. After sticking alarm, dynamic control strategies of casting speed according to the location of alarmed thermocouples are proposed for breakout prevention, by investigation on formation mechanism and recovery behavior of sticking-type breakout. The proposed model and control strategy have been validated through its application on a steel plant. The results show that the detection ratio for stickers and frequency of false alarm are 100% and 0.15% times/heat respectively, and the healing rate of the stickers has reached 100%. The case in which breakout still occurs after alarm has been eliminated. The present technology has achieved better performance in prediction and control of the sticking-type breakout and satisfies the requirement for field application. (C) 2015 Elsevier Ltd. All rights reserved.