二方向カメラ画像を用いた傾動式取鍋注湯流量の高精度非接触計測

抄録

  This paper proposes an image-based, non-contact method to measure the pouring flow rate in a tilting-ladle process. Pouring is a key process affecting product quality in foundry production. While conventional load-cell-based estimation has low sensitivity in short pouring processes because the variation of molten-metal mass is small relative to the ladle’s total mass, the proposed method estimates the instantaneous flow rate from camera images by simultaneously computing a representative velocity and falling liquid cross-sectional area (hereafter referred to as the falling liquid). In the proposed approach, two cameras are placed at the front and side views to detect the outline of the falling liquid in real time. In hydrostatic-tank experiments, the proposed vision-based IAE was comparable to that of the clamp-on flowmeter in overall magnitude, while enabling fully non-contact measurements. Robot pouring tests (robotic tilting pouring experiments) have demonstrated that the method has practical feasibility, achieving a mass estimation error rate of －1.47%. These results indicate that the proposed method provides a non-contact sensing method applicable to automated pouring.