Research On Ball Mill Load Prediction Method Based On Multi-source Signal Fusion Technology

Ball mill has the advantages of simple operation,low manufacturing cost and large crushing ratio.It can be used in wet milling and dry milling.It is widely used in glass,ceramics,cement,chemical industry,mining and other fields.However,there are many shortcomings in the milling process of ball mill,such as the mutual limitation,strong coupling and long lagging time and so on.These shortcoming cause that the internal load parameters of the cylinder can not be described and controlled in time,so it is difficult to give full play to the actual performance of the mill.Therefore,to achieve effective prediction of the internal load of the mill,so that the ball mill running in the best working condition,is one of the fundamental tasks to improve the milling efficiency,reduce production costs.In this paper,the experimental ball mill as the research object,through the combined method of experience analysis,experimental research,signal processing,use multi-sensor to detect the ball mill bearing vibration signal,cylinder mill sound signal and the main motor current signal.Based on multi-source information fusion technology of optimal fusion and D-S evidence theory,the method of multi-source signal feature extraction and prediction for mill load is studied in this paper.The effective prediction of the internal load parameters of the Ball mill is realized.The main results were as follows:First,the main influence factors of grinding process and a signal detection method obtained through empirical analysis.The multi-source signal detection system of the ball mill is built and the single factor variable method is used to calculate the grinding medium.The load parameters of different grinding machines can be divided into three groups : The evaluation index is that the energy consumption and-200 mesh yield,and the energy consumption,the feed volume,the feed particle size distribution and the ball ratio as the input parameters.Second,for the characteristics of multi-source signal extraction and recognition of three kinds of state,using the wavelet transform technique in ball mill vibration,grinding sound signal feature extraction.The mean value,variance and energy value of vibration signal are obtained and mill sound characteristic information signal A weighted sound pressure level and A weighted octave sound pressure level.By comparing and analyzing the Euclidean distance of signal characteristic values under different working conditions,the results show that the multi-source signal can be more accurate and faster than the single signal.The averageprocessing of the current signal in different time periods,and the current value increases first and then decreases with the increase of the load.Finally,Multi-source signal of the mill load forecasting,which is high conflict,strong mutation and low correlation,The improved optimal fusion algorithm is adopted,The detection data of similar signals in different time periods are fused,The results show that this method can effectively eliminate the high conflict information;Using the improved D-S evidence theory rule,A multi source heterogeneous signal feature fusion method for mill load is established,Through the example verification and comparison of different algorithms,It shows that the fusion result is higher confidence,faster convergence and better reliability when this method is applied to mill load forecasting.In summary,Through single factor variable grinding test and multi-source signal feature extraction and recognition,Fusion method of multi-source signal feature layer based on optimal fusion and D-S evidence theory,It has strong practicability and reliability for the load prediction of the mill,and it can also provide a new idea for the energy saving and consumption reduction of other mineral processing equipment.