多次元情報抽象化レベルを有するキュービックニューラルネットワークによる流動床式ごみ焼却炉の知的制御

概要

論文の詳細を見る
Fluidized bed incinerators can burn waste in a short time by stirring waste and fluidized sand. However, it is difficult to stabilize an incinetor plant without operators in abnormal situations and to obtain exact mathematical models. Moreover they tend to change their characteristics. Thus, it is necessary for the plant control to realize an intelligent control method which stabilizes plants when abnormal situations occur, which are not only expected but also unexpected and when their characteristics change. As a solution to such a control problem, we present a method of intelligent control using "cubic neural network" (CNN) which possesses multi-levels of different degree of abstracted signals and is capable of parallel distributed signal processing. As an unexpected abnormal situation, we consider sensor failure. The control ability of this method was verified by experiments on an actual plant.
一般社団法人日本機械学会の論文
1997-01-25