Adaptive sigmoidal molten metal pouring control

We present a new adaptive nonlinear controller for vision-based molten metal automatic pouring. We describe the challenges, modeling, identification, and control of the process. Attempts to employ proportional integral (PI) and proportional integral derivative (PID) controllers were partially successful. An adaptive sigmoidal controller improved the control quality due to its variable gain and bias. The design has been successfully implemented by the Inductotherm Corp. in a new automatic pouring system named VISIPOUR