105 ニューラルネットワークを用いた粗大ごみ破砕処理の自動化

Abstract

This paper describes an automatic operating system with Neural Networks for a bulky-refuse shredding plant. One of the important technologies in the automatic operating system is the measurement, with an image processing unit, of the volume of raw materials. However, it is sometimes impossible to measure the material volume correctly only with the image processing unit, for example, in a dusty environment. As one of the solutions, we applied Neural Networks for the volume estimation, and have verified the advantage of using Neural Networks through a series of validation tests in a real shredding plant.