Abstract: Control precision of coiling temperature is one of the key factors influencing on the strip quality/shape during the cooling process of hot rolled steel strip. The temperature control precision is therefore the core of the mathematical model to control cooling process and, at the same time, the mathematical model should serve the function of self-learning to improve the temperature control precision. A mathematical model with nonlinear structural characteristics was developed to control the cooling process. Analyzing its self-learning ability, the model enables the key parameters to be corrected uninterruptedly so as to improve the temperature control precision and adaptability of the model. The in-situ application results verified that the mathematical model developed can provide high control precision of coiling temperature, thus laying a foundation for improving steel strip quality.