Grade prediction of zinc tailings using an encoder-decoder model in froth flotation

•An encoder-decoder-based network predicts the zinc tailings grade in advance.•The feature time series of the first rougher is extracted automatically.•The proposed grade prediction model considers previously measured grades.•The dynamic consistency between the input and output time series is utiliz...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	Minerals engineering Ročník 172; s. 107173
Hlavní autoři:	Zhang, Hu, Tang, Zhaohui, Xie, Yongfang, Luo, Jin, Chen, Qing, Gui, Weihua
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	Elsevier Ltd 01.10.2021
Témata:	Encoder-decoder model Froth video Grade prediction Machine vision Recurrent neural network Time series Encoder-decoder model Machine vision Recurrent neural network Grade prediction Froth video Time series
ISSN:	0892-6875, 1872-9444
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Popis
Shrnutí:	•An encoder-decoder-based network predicts the zinc tailings grade in advance.•The feature time series of the first rougher is extracted automatically.•The proposed grade prediction model considers previously measured grades.•The dynamic consistency between the input and output time series is utilized.•The proposed grade prediction model was validated in a zinc flotation process. Accurate grade prediction is conducive to proper flotation operation or control. Different from grade monitoring, grade prediction needs to obtain the target grade in advance. However, there is usually a time delay between the flotation cell and the predicted grade in froth flotation. This time delay makes it difficult to match the features of the observed flotation cell with the predicted grade. To solve this problem, this article studies the method of zinc tailings grade prediction using encoder-decoder models. The proposed model considers the feature time series of the first rougher and the previously measured tailings grades. First, according to the sample ratio between froth video and X-ray fluorescence (XRF) analyser, the feature time series of the first rougher can be automatically extracted by finding the nearest available feature vectors. Next, the feature time series of the first rougher is fed into the encoder to generate a context vector, and then the context vector and previously measured grades are sent into the decoder to predict the current tailings grade. The proposed model effectively captures the dynamic consistency between the feature time series and previously measured grades. The effectiveness of the proposed model in the froth flotation has been verified by experiments. Compared with the traditional recurrent neural network (RNN)-based models, the root mean squared error (RMSE) and mean absolute percentage error (MAPE) of the proposed model decrease by about 17.8% and 1.9%. respectively, and the R-squared (R2) score of the proposed model increases by about 13.8%.
ISSN:	0892-6875 1872-9444
DOI:	10.1016/j.mineng.2021.107173