Online prediction of concentrate grade in flotation process based on PCA and improved BP neural networks

摘要

According to the difficulty of online measure of concentrate grade during mineral flotation process, an online prediction method for concentrate grade based on PCA and improved BP neural networks is proposed. Firstly, bubble characteristics are extracted from real-time obtained images by means of digital image process technology and their relationships to concentrate grade are analyzed. Secondly, some principal components are extracted through PCA algorithm from these characteristics. Finally, an improved BP neural networks algorithm is adopted to construct prediction model which takes the concentrate grade data collected by offline assay as the training objectives. The experimental results demonstrate that the proposed method can effectively predict flotation concentrate grade.