17 April 2024 in IEEE Transactions on Industrial Informatics
The process operating performance assessment (POPA) of froth flotation under multimode is an important means to improve production efficiency and product quality. In the plant, froth flotation presents multimode characteristics due to the uncontrollable of the objective conditions of production. In this article, a method based on graph representation learning is studied to solve the POPA problem of a multimode froth flotation process. Based on the qualitative and quantitative information represented by the froth images and process variables of the flotation process, an appropriate convolutional neural network is selected to fuse the multisource heterogeneous information. The powerful representation capabilities of deep clustering methods are used to divide working modes. A deep embedded graph clustering network is proposed to learn the node representation, on which the clustering effect of the model is improved by reconstructing the topological structure and node content of the graph data. In addition, a dual self-monitoring mechanism is designed to guide the model parameter update. Finally, a soft recognition method is designed to fuse the performance grades under different modes by using Bayesian fusion theory, which improves the accuracy of assessment. The proposed method is applied to a gold flotation process to illustrate the feasibility of the method.
30 August 2022 in IEEE Transactions on Industrial Informatics
The process operation performance assessment in the smelting process of electro-fused magnesium furnace (EFMF) plays a very important role in improving production efficiency and economic benefits. Most of the existing methods mainly rely on a large number of labeled data, which is difficult to satisfy in practical application. To solve the problem of a small number of labeled samples, a semi-supervised transfer generative adversarial network (SSTGAN) based on source domain (SD) data is proposed to assess the operation performance of EFMF. The core idea is to use the labeled data of SD to train a classifier and predict the unlabeled data of the target domain (TD), and use the GAN to make the multi-source heterogeneous information comprehensive features of the TD and SD conform to the similar distribution. The experiment shows that the method can obtain satisfactory assessment results than the compared method.
12 December 2022 in IEEE Transactions on Instrumentation and Measurement
Operating performance assessment of gold flotation process plays an important role in improving the metallurgical performances and pursue the best comprehensive economic benefits. The appearance of froth is a good indicator of flotation performance, however, labeling the data is costly. To assess the operational performance of flotation process with a small amount of labeled data, this study proposes a semi-supervised cluster generative adversarial network (SSClusterGAN). First, latent variables are sampled from one-hot encoded variables and continuous normal distribution variables. While projecting all data into the latent variable space, the specific types of clustering in the latent space is realized by controlling the type of distribution of the learned latent codes, so that the training objectives of labeled data and unlabeled data are consistent. Then, by adapting the generative adversarial networks (GAN) framework, a pair of stacked discriminators are used to learn the conditional distribution of attributes for labeled and unlabeled data, respectively. In addition, using semi-supervised learning (SSL) as the basic learning technology, the high-confidence pseudo-labels obtained by training with unlabeled data are combined with the labeled training data to increase the training samples and effectively improve the accuracy of the model. Finally, the method was applied to the gold flotation process. The experimental results show that the method effectively utilizes unlabeled samples to expand the labeled dataset, combines the advantages of semi-supervised learning and GAN, obtains accurate assessment results, and effectively improves the comprehensive economic benefits of the gold flotation process.
30 January 2023 in IEEE Transactions on Instrumentation and Measurement
The process operation performance assessment (POPA) of electro-fused magnesium furnace (EFMF) plays an irreplaceable role in improving production efficiency and ensuring product quality. The existing methods of POPA seriously rely on a large number of manually marked label data. However, manual marking of all samples will pay a very high price, which is not applicable in the actual industrial application of EFMF. Therefore, taking advantage of the multi-source heterogeneous information of image, sound and current generated by EFMF, we propose a novel semi-supervised fine-grained generative adversarial network (SSFGGAN) that addresses the issue of POPA of EFMF smelting process. The SSFGGAN method makes use of generative adversarial network and semi-supervised strategy, and works together with the classifier, which makes the feature distribution of each operation performance grade in source domain (SD) and target domain (TD) consistent through the adversarial learning of a fine-grained discriminator. Therefore, the comprehensive features learned in SD remain effective for TD data, and can be directly applied to POPA of TD. Finally, the simulation results show that the assessment accuracy has been improved with a small number of labeled data and a large number of unlabeled data. The accuracy of SSFGGAN method can reach 99.875%, 99.75%, 99.625% and 99.125% respectively when the ratio of labeled data to unlabeled data is 1:2, 1:5, 1:10 and 1:20. In each evaluation index, it is better than the compared methods.
4 January 2024 in IEEE Transactions on Industrial Informatics
In the process operation performance assessment (POPA) of electro-fused magnesium furnace (EFMF), the distribution of testing and training data is inconsistent. Therefore, in order to solve the problem, a new adaptation entropy regularization actor–critic (AERAC) reinforcement learning method is proposed based on the multisource heterogeneous information. The core of the method is that the parameters $\theta _{\text{history}}$ of the POPA model trained with historical data and the POPA parameters $\theta _{\text{current}}$ trained with current data are combined into the state input policy network. Then, the policy network predicts the action according to the state. Finally, the action will integrate the parameters $\theta _{\text{history}}$ and $\theta _{\text{current}}$ into new parameters $\theta _{\text{test}}$ of the POPA model for the adaptive test dataset. The method is verified by the dataset of EFMF. The results show that the AERAC method can effectively solve the problem that the distributions of training and testing set are different.