Deteksi Cacat Plate Insulator Menggunakan Backpropagation Neural Network

Abstract

Visual Inspection Plate Insulator Check is a critical step of quality control to ensure materials used are pass quality standard. The problem of manual material inspection can not detect whether the materials used for production are fine (OK) or not good (NG). Because of the failure in the inspection, there will be material and manpower loss due to rejection. In addition, it may make more problems if this defect product (NG) is delivered to customer. Then, it will jeropardize the company credibility. The problem of inspection failure can happen because of human factor such as tired, inaccurate, low-skill operator inspection with visual check. Therefore in this study, the researcher will develop a visual inspection technique of material checking with imaging  processing backpropagation neural network. Using image processing technique, the differences between OK materials and NG material will be investigated from the both images. The product images will be captured in production line and the comparison will be observed with the OK product. It is expected, this inspection will be more accurate and does not depend on human factors. Finally, the technique can be developed further into automatic visual inspection by material checking, e.g. a kind of Plastic.