Data-driven control knowledge acquisition and online updating method for grinding classification process
ZHOU Jia-yi
SUN Kai-xuan
WANG Xiao-li
YANG Chun-hua
ZOU Mei-yin
Abstract：The mechanism of grinding process is complex and the flow is long.The traditional optimal control method has weak self-learning ability,and it is difficult to make effective control response in long-time under the frequent changes of complex ore sources and operating conditions.Therefore,a strategy for online extracting and updating the control knowl-edge of grinding-classification process is proposed in this paper.Firstly,the operating conditions of grinding classification process are divided finely based on state transition algorithm-based kernel fuzzy C-means clustering method.The optimal operating conditions are then selected by combining the process operating characteristics and clustering results.Next,a Wang Mendel algorithm based on weighted optimization is proposed to extract the optimal control knowledge of different operating conditions and rule confidence is defined to evaluate the effectiveness of the knowledge.Finally,the control rules are updated online based on the double sliding window mechanism.The results show that compared with control using offline rules,adaptive fuzzy logic control and manual control,control using online rules has better adaptive ability.
Keywords：grinding-classification processoptimal controlknowledge acquisitionknowledge online updating
Publication Date：2025-02-28
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：9( 217-225 )