A Soft Sensing Method for Mill Load Parameters Based on Virtual Samples

Abstract

The invention discloses a soft measurement method of mill load parameters based on virtual samples. The method first uses integrated empirical mode decomposition technology (EEMD) to obtain multi-scale time domain sub-signals of mill cylinder vibration and vibroacoustic sample signals. , after further processing, high-dimensional spectral data with different time scales are obtained; then based on these high-dimensional spectral data, the improved selective integrated kernel partial least squares method (IGASEN‑KPLS) is used to build a feasibility-based planning (FBP) model , and generate new virtual samples based on prior knowledge and the FBP model; then mix them with real training samples to obtain hybrid modeling samples, and use a feature selection method based on mutual information (MI) to adapt multi-scale spectral features Select, use these selected spectral features to build a soft sensor model, and perform soft sensor measurements.

Description

technical field [0001] The invention relates to the field of soft measurement, in particular to a method for soft measurement of mill load parameters based on virtual samples. Background technique [0002] The optimal operation control of the grinding process needs to accurately detect the load parameters in the mill (see literature [1] P.Zhou, T.Y.Chai, H.Wang, "Intelligent optimal-setting control for grinding circuits of mineral processing," IEEE Transactions on Automation Science and Engineering, 6(2009) 730-743. and literature [2] T.Y.Chai, "Operational optimization and feedback control for complex industrial processes," Acta Automatica Sinica, 39(2013) 1744-1757). Tens of thousands of steel balls are arranged in layers inside the mill, and the impact forces of different layers of steel balls on the materials inside the mill and the barrel of the mill have different strengths and periods. Usually the measured cylinder vibration signal is a mixture of multiple sub-signal...

AI Technical Summary

Problems solved by technology

However, due to the continuous and closed operation of the ball mill, the modeling data used to construct the soft-sensing model can only be obtained in the experimental design stage for the construction of the soft-sensing model or in the restarting operation stage after the production process is stopped, and sufficient effective data can be obtained. The economic and time cost of modeling data is high

Images