本文描述了神经网络在柔顺刀具去毛刺操作中的应用。实验是用研磨浸渍尼龙丝状刷进行的，以获得有关刷力响应和毛刺高度减少的信息。开发了两种不同的神经网络，并分别使用力响应和毛刺高度减少数据进行训练。然后，这些网络预测刷力响应和毛刺高度降低。我们观察到，经过训练的神经网络是一个有效的预测工具，可以预测刷力响应和使用研磨浸渍尼龙丝状刷获得的毛刺高度降低。

THIS PAPER DESCRIBES THE APPLICATION OF NEURAL NETWORKS TO A COMPLIANT TOOL DEBURRING OPERATION. EXPERIMENTS ARE CONDUCTED WITH AN ABRASIVE IMPREGNATED NYLON FILAMENTARY BRUSH TO OBTAIN INFORMATION ON BRUSH FORCE RESPONSE AND BURR HEIGHT REDUCTION. TWO DIFFERENT NEURAL NETWORKS ARE DEVELOPED AND TRAINING WITH FORCE RESPONSE AND BURR HEIGHT REDUCTION DATA, RESPECTIVELY. THESE NETWORKS THEN PREDICT BRUSH FORCE RESPONSE AND BURR HEIGHT REDUCTION. IT IS OBSERVED THAT A TRAINED NEURAL NETWORK IS A VALID PREDICTOR OF BOTH BRUSH FORCE RESPONSE AND BURR HEIGHT REDUCTION OBTAINED BY USING AN ABRASIVE IMPREGNATED NYLON FILAMENTARY BRUSH.