Research On Load Prediction Of Ball Mill Based On Feature Extraction Of Grinding Sound Signal

Ball mill is an important industrial production equipment.It is widely used in cement,glass,metal beneficiation and other production industries.It can grind various materials and ores into fine particles.The working load of the ball mill is dynamically changing.At present,the working load of most ball mills is not detected,and the corresponding control system does not consider the change of the working load,and it is difficult to achieve optimal control.When the load of the ball mill is different,its audio signal characteristic changes.By detecting the characteristic value of the audio signal,the change of the working load of the ball mill can be judged.This paper studies the audio characteristics of the working load of the ball mill,seeks the relationship between the changing state of the working load and the corresponding audio characteristics,and makes basic research for the optimal control of the ball mill.This article takes the laboratory small ball mill as the research object,through the combination of theoretical research,experimental exploration and simulation analysis.First of all,starting from the structure and working principle of the ball mill,the mechanism of grinding sound is analyzed to determine the research focus of this article.The mill sound signal acquisition system of the ball mill was built,the experimental scheme was designed,and the mill sound signals of the ball mill under different loads were collected through experiments,which provided external characteristic signals for the load prediction of the ball mill.Secondly,the power spectrum of the grinding sound signal is estimated.Through analysis,the spectrum of the grinding sound signal changes with the change of load,and the change of the power spectrum of the grinding signal of the ball mill has the characteristics of randomness,complexity and nonlinearity;Using the principal component analysis method to extract the basic features after segmenting the power spectrum of the ball mill's grinding sound,and then through analysis,we know that the feature extraction reduces the dimension of the features and the redundancy between the features,making the feature distribution more deterministic and random Sex is smaller.Finally,the parameters of the support vector machine are analyzed and researched.It is proposed to use the particle swarm algorithm to optimize the parameters of the support vector machine.The optimized parameters are used to establish the support vector machine prediction model.The sample set after feature extraction is used to train the model and Make predictions,and compare the prediction results with the prediction results of the two models of PCA-SVM and SVM.The results show that the PCA-PSO-SVM prediction model proposed in this paper can effectively predict the load of the ball mill,and has high prediction accuracy and stability.It has the advantages of good performance and high calculation efficiency.Through the above research,it has been shown that grinding sound signals collected by grinding experiments,feature extraction of grinding sound signals,and the establishment of a support vector machine prediction model based on particle swarm optimization algorithm have achieved effective prediction of ball mill load,which has improved the efficiency of ball mill grinding and reduced Energy consumption has a positive guiding role.