In this study, a method for realizing an intelligent production process by reducing quality variation in the manufacturing industry is proposed. A quality fluctuation model in a production process is abstracted, and the quality is improved using adaptation rules based on the model. In this framework, the value directly related to product quality is expressed as multiplying the coefficient and the setting parameter. This expression makes it possible to regard the quality variations as being caused by the coefficient variations. Hence, it is possible to reduce the variation of quality by predicting the fluctuation of the coefficient from various data acquired from the production line and increasing or decreasing the setting parameter based on the predicted value. Moreover, the element description method is applied to predict the fluctuation of the coefficient. The element description method has the advantages of a model-based method whose physical meaning can be understood and the advantages of a database method applicable to an unknown system. Therefore, the mechanism of fluctuation can be abstracted and can be used as explicit knowledge. In this study, this framework is applied to reduce the variation in filling weight of the powder filling process and is demonstrated. As a result, the filling weight variation has been reduced by approximately 33%.
|ジャーナル||IEEE Open Journal of the Industrial Electronics Society|
|出版ステータス||Published - 2021|
ASJC Scopus subject areas