Abstract: The cooling process of rolled plate only depend on feedforward model, but the existing precision of commonly used temperature models are not high enough for controlling properly the process. A differential equation was therefore deduced as the model to control the plate cooling process, of which the 1st and 2nd, especially the self-learning modifications were carried out for calculation in combination with the reform of the cooling control system for Shougang's Plate Mill, thus providing the precision systematically for its temperature control. A calculation using a heat transfer coefficient is presented in plate laminar cooling to adapt to the plate cooling characteristics, then a selflearning algorithm of laminar flow cooling is given, by which the model's control precision is further improved. The control cooling system has been applied to Shougang's Plate Mill and showed a 96.5% acceptability in temperature control.