The invention relates to the technical field of thin-strip
continuous casting production, in particular to a method for automatically controlling the height of a
molten pool in the double-roller thin-strip
continuous casting production process, which comprises the following steps: arranging a
laser ranging module and a
visual perception module;
processing by an
embedded processing module; establishing a
molten pool height
dynamic prediction model by adopting a
deep learning algorithm, and outputting a
molten pool height prediction value; self-adaptive
sliding mode control is adopted as a main controller, and the displacement amount of the
servo hydraulic stopper rod is output; and controlling the height of the molten
pool based on the displacement of the
servo hydraulic stopper rod. According to the method,
laser is used for providing accurate distance reference, visual sensing is used for providing molten
pool liquid level information, and the detection precision is improved and environmental interference is reduced through a
machine vision
algorithm; based on the process parameters and the real-
time information of the sensor, constructing a molten
pool height prediction model by using a neural network
algorithm with physical constraints; and a
reinforcement learning optimizer is added to adjust sliding-
mode control parameters online, so that the sliding-
mode control parameters adapt to working condition changes to carry out high-precision and fast-reaction molten pool height control.