Grade Monitoring using Semantic Features of Flotation Froth Image

Abstract

In the industrial flotation circuits, grades are the key performance indicators for flotation condition recognition and operations. Modeling based on machine vision is an effective tool for grade monitoring, and froth image feature extraction as a model input plays a crucial role. Given the low efficiency of hand-crafted features and the poor interpretability of high-dimensional features extracted by convolutional neural networks, a grade monitoring method with human-understandable semantic features as input is proposed in this paper. First, based on the combination of generative adversarial networks and encoder, we design a new network structure that maps the froth images data space to the latent semantic space. Then the semantic features of froth images can be automatically extracted by decomposing the constructed matrix with rich latent semantics. Finally, as demonstrated in the industrial experiment, the extracted semantic features can not only be visually interpreted but also can be effectively used in grade monitoring.