Related Experiment Video
Updated: Apr 27, 2026
Extraction and Detection of Geosmin and 2-Methylisoborneol in Water and Fish using High-Capacity Sorptive Extraction Probes and GC-MS
Published on: July 3, 2025
Improved GSO optimized ESN soft-sensor model of flotation process based on multisource heterogeneous information
Jie-sheng Wang1, Shuang Han2, Na-na Shen2
1School of Electronic and Information Engineering, University of Science & Technology Liaoning, Anshan 114044, China ; National Financial Security and System Equipment Engineering Research Center, University of Science & Technology Liaoning, Anshan 114044, China.
This study introduces an improved echo state network (ESN) model for predicting flotation process indicators. The novel approach enhances prediction accuracy for real-time control in mineral processing.
Area of Science:
- Mineral Processing
- Artificial Intelligence
- Process Control
Background:
- Flotation is a crucial process in mineral separation.
- Accurate prediction of key indicators like concentrate grade and tailings recovery is essential for process optimization.
- Existing soft-sensor models often face challenges with high-dimensional data and prediction accuracy.
Purpose of the Study:
- To develop a robust soft-sensor model for predicting flotation process indicators.
- To enhance the prediction accuracy and generalization capability of echo state networks (ESNs) for real-time applications.
- To integrate image processing techniques with advanced machine learning for improved process monitoring.
Main Methods:
- Utilizing color (saturation, brightness) and texture features (GLCM-based) from flotation froth images.
- Applying kernel principal component analysis (KPCA) for dimensionality reduction of input data.
- Developing an echo state network (ESN) based fusion soft-sensor model.
- Optimizing the ESN model using an improved glowworm swarm optimization (GSO) algorithm with a congestion factor.
Main Results:
- The proposed ESN soft-sensor model demonstrated superior prediction accuracy compared to traditional methods.
- The fusion of image features and process data improved the model's ability to capture complex process dynamics.
- KPCA effectively reduced input dimensionality and ESN complexity, enhancing computational efficiency.
- The optimized ESN model met the requirements for real-time control in the flotation process.
Conclusions:
- The developed ESN-based fusion soft-sensor model, optimized with GSO, is effective for predicting key flotation indicators.
- The integration of image analysis and advanced machine learning offers a promising approach for intelligent mineral processing.
- The model's enhanced generalization and prediction accuracy support its application in real-time process control and optimization.
