Reaction Force Inspection System Using Neural Network Classifier

Abstract

People recognize the quality of a product while in operation by hands or fingers. The operation feeling by hands or fingers is one of the important indexes for the high-grade products. However, skilled inspectors are used to inspect some products because automatic inspection is technologically difficult or too high in cost. This paper looks at a system for inspection of the quality of a product’s reaction force characteristics. This system, until now considered difficult to realize, automates the inspection method utilizing the touching of an inspector's finger. Neural network classifier is applied to the system for products to learn an inspector's finger judgment. We provide an input layer of a neural network classifier with nodes corresponding to time-and frequency-domain features of reaction forces of a product and an output layer with three nodes corresponding to a judgment; being one of non-defective, defective, or unable to judge. From experimental results, the effectiveness of the proposed neural network classifier has been clarified.