Soft Sensor Modeling of Mill Load Based on Feature Selection Using Synergy Interval Pls

Mill load is an important equipment index which is closely related to operating efficiency, product quality and energy consumption of grinding process. Due to high dimension and collinearity of spectral data, mill load model has high complexity, poor interpretability and generalization. A soft sensor modeling method of mill load parameters is proposed based on frequency spectrum feature using Synergy Interval Partial Least-Squares Regression (SiPLS). Based on the spectrum feature of the shell vibration or acoustic signal, three soft sensor models of mill load, such as mineral to ball volume ratio, charge volume ratio and pulp density are developed, respectively. The proposed method is tested by the wet ball mill in the laboratory grinding process. The experimental results have demonstrated the proposed method has higher accuracy and better generalization performance than the full-spectrum model and iPLS feature spectrum model, and the feature spectrum model based on the shell vibration is superior to the acoustic feature spectrum model.