LAUSR

Abstract Based on machine vision and image processing technology, a large number of methods have been proposed for flotation working condition recognition in froth flotation industry. However, in these condition recognition methods, there lacks internal or external evaluation criteria, so that the reliability of the recognition results obtained by the current recognition models is not high for operators. In order to solve this problem, by taking time series of working condition data as the research object, an improved Apriori algorithm based on time series is proposed for the evaluation of flotation working condition recognition. Based on the idea of transaction database, the time series data set of working condition is divided into working condition transaction database by an auto-regressive model. Then the improved Apriori algorithm is used to find the association rules that contain relations between the time-series of the working condition transaction database to form a rule base for the working condition, which provides the reliability evaluation of the online working condition recognition results based on the association rules.