基于小波神经网络和DEM试验的球磨机钢球动能预测

Prediction of kinematic energy of grinding balls in ball mill based on wavelet neural network and DEM test

球磨机内存煤量是制粉系统里磨煤控制量中最重要的控制因素之一。为探索球磨机存煤量与钢球动能之间的重要关系,通过基于离散元素法的PFC3D软件平台模拟球磨机运行过程,随着球磨机内煤量的不断增加,考察固定球径时,钢球的实时动能和体现磨机效率的球荷摩擦损失能量与筒壁做功的对比变化规律,并得出实时数据。基于该实时数据,采用小波神经网络方法,通过球磨机内存煤量来预测钢球的运动动能。预测结果表明,小波神经网络用于预测球磨机内的钢球动能可行,预测的钢球动能与实际动能有着相似的变化规律和很小的预测误差,并且比BP神经网络的预测结果更加准确;证明了球磨机存煤量和钢球动能之间存在密切联系,为今后用钢球动能来控制球磨机内存煤量的研究提供了理论基础。

The coal storage in the ball mill is one of the most important control factors for coal pulverization in pulverization system. In order to explore the significant relationship between coal storage and kinematic energy of grinding balls, PFC3 D software platform based on discrete element method was applied to simulate the operation of the ball mill. With the growth of coal storage in the ball mill, at the fixed diameter of grinding balls, variation laws of real-time kinematic energy, energy loss due to balls friction and work done by the shell wall which reflected the mill efficiency were observed, and real-time data were obtained. Based on the real-time data, the wavelet neural network was applied to predict the kinematic energy of grinding balls by coal storage in the ball mill. The prediction results showed: the prediction of the kinematic energy of grinding balls in the ball mill by the wavelet neural network was feasible, and predicted kinematic energy had the similar variation laws with actual one; there was only a litter error between the two, and the predicted energy was more precise than the one obtained by BP neural network. In addition, close relation between coal storage in the ball mill and kinematic energy of grinding balls was confirmed, which offered theoretical foundation for the study on the control of coal storage in the ball mill by kinematic energy of grinding balls.