Abstract
With the development of the new technology of intelligent manufacturing and cyber physical system, a new scheme is proposed for designing of predictive production based on ball mill. First, physical model (PM) and the model based on data (cyber model, CM) are discussed. Then, the combination of physical model and cyber model (CPM) is realized. Physical model is established according to the volume balance formula and the material balance formula. Cyber model uses the algorithm of the extreme learning machine which introduces penalty function. CPM uses the least square method to realize the combination of PM and CM and gets the value of the coefficient. Compare the actual data on ball mill to the data of the model then the result shows that the mean square error of CPM is smaller than the mean square error of PM and CM. The experimental results validate the effectiveness of the proposed method, which can be effectively used in ball mill in our industry.