Zinc flotation dosing state evaluation method based on probabilistic semantic analysis model

Abstract

The invention discloses a zinc flotation dosing state evaluation method based on a probabilistic semantic analysis model. The method comprises the following steps of obtaining dynamic characteristics of the zinc flotation images in different dosing states through an acceleration robust characteristic algorithm; extracting bottom layer texture features of the foam image by utilizing a gray level co-occurrence matrix; clustering the obtained dynamic feature vectors of the foam images through an accurate Euclidean local sensitive hash clustering algorithm; constructing a relevant visual dictionary, measuring the correlation degree between the obtained dynamic features and the obtained bottom texture features by using Pearson correlation coefficients, optimizing the original visual dictionary, and finally, carrying out zinc flotation dosing state evaluation on different foam images by using a probability semantic analysis model. The problems of inaccurate evaluation and long time consumption of the zinc flotation dosing state by workers are solved, the zinc flotation dosing state can be accurately evaluated, the operation time is shortened, and therefore the overall optimization of the zinc flotation process is achieved.

Description

technical field [0001] The invention relates to the technical field of zinc flotation automation, in particular to a zinc flotation dosing state evaluation method based on a probabilistic semantic analysis model. Background technique [0002] In the field of zinc flotation, the evaluation of dosing status is mainly done by experienced workers on site observing the foam morphology on the surface of the flotation tank to complete the on-site operation. The operation error of the dosing state is large and the efficiency is low, and it is impossible to objectively evaluate and recognize the dosing state in the zinc flotation process, which easily leads to low work efficiency and loss of mineral raw materials. [0003] In the field of zinc flotation dosing status evaluation, it was mainly identified by adaptively learning the surface characteristics of the bubbles. By converting the PDF of the bubble size into a cumulative distribution histogram feature, the typical reagents were...

AI Technical Summary

Problems solved by technology

[0002] In the field of zinc flotation, the evaluation of dosing status is mainly done by experienced workers on site observing the foam morphology on the surface of the flotation tank to complete the on-site operation. The operation error of the dosing state is large and the efficiency is low, and it is impossible to objectively evaluate and recognize the dosing state in the zinc flotation process, which easily leads to low work efficiency and loss of mineral raw materials

[0003] In the field of zinc flotation dosing status evaluation, it was mainly identified by adaptively learning the surface characteristics of the bubbles. By converting the PDF of the bubble size into a cumulative distribution histogram feature, the typical reagents were obtained by using the unsupervised furthest neighbor clustering learning method. Under the clustering feature set of the cumulative distribution histogram of foam size, the typical distribution of a section of foam image is marked and the Bayesian reasoning principle is used to obtain the evaluation of the dosing state under the current dosage. This type of method mainly has complex calculations and evaluation inaccurate deficiency

Images