It is difficult to effectively control the vertical grinding process of raw materials due to its characteristics of strong coupling, non-linearity and large hysteresis. This paper proposes a vertical mill intelligent control system based on data mining to predict the operating conditions of the slag grinding system. Taken into consideration corresponding shortcomings of each algorithm, we combine several algorithms to propose a feature extraction method for analyzing operating conditions and determining the indicators that affect the operation. Next, we clustered the healthy operating conditions to get the distribution of health conditions, and based on this, established a healthy operating condition library. The operational data are compared with the reference conditions, and the prediction model is trained using the ARIMA algorithm to predict the trend of the corresponding indicators. To verify the effectiveness and practicability of the method, we developed a software system and applied it to the actual case analysis. It is concluded that the vibration of the control group is decreased by an average of 10%, and the average power consumption per ton is decreased by 6.05%. According to the total number of vertical mills of 350,000 tons, the average power consumption per ton is 43.5 degrees. Therefore, the total annual power consumption will be 1.5225 million kilowatt hours, which can save 921,100 kilowatt hours. According to the average industrial price of 1.5 yuan / kWh, the annual saving will be 1,381,700 yuan.
Keywords Data mining, Vertical mill, Health operating conditions, Clustering analysis, Intelligent control