Soft sensing of mill load parameters based on multi-scale frequency spectrum

Soundly shell vibration and acoustical signals produced by ball mill of grinding process contain useful information for judging parameters inside ball mill. These signals consist of different time-scale sub-signals which are caused by different reasons and have different physical interpretation. In this paper, a new multi-scale frequency spectrum feature selection and extraction based on soft sensing approach is proposed to estimate the load parameters of wet ball mill. This approach can extract and select different scales' frequency spectrum features. In this study, the mill shell vibration and acoustical signals are first decomposed into multi-scale time domain sub-signals by empirical mode decomposition (EMD). Then multi-scale frequency spectrums are obtained by fast Fourier transform (FFT) to these sub-signals. Thirdly, spectral principal components and characteristic frequency sub-bands are extracted and selected from the multi-scale frequency spectrum. Finally, a combinatorial optimization method selects the input sub-set and parameters of the soft sensor model simultaneously. This approach is successfully applied in a laboratory scale wet ball mill. The test results show that the proposed approach is effective for modeling parameters of mill load.