Soft Sensor For Ball Mill Level Based On T-S Fuzzy Model

Ball mill is a common large machine for grinding materials in industrial manufacturer. During practical production process, steel balls and materials hit each other frequently, which made ball mill hard to keep working in stable state and brought difficulties in controlling effectively. The data of the research shows that the grinding process of ball mill is mainly self-consumption. Ball mill is high energy-consuming and low-efficient equipment which can’t often. So it is extremely important to conduct the best control of the load of ball mill for energy-saving and cost-reducing and steady running. But the most prominent and difficult problem is the effective and precise measurement of the load (the material level) in the control of the mill load. Because in the process of grinding, the steel ball and materials deformed physically as the roller’s rolling, which cause more uncertainty, like variable and complex environment inside the roller. It brings enormous difficulty in measuring the level directly and precisely. And if the traditional testing method is adopted, it may cause low accuracy and poor stability.In the paper, the operation mechanism of ball mill is analyzed, and the bearing vibration signals were collected and processed. The vibration feature parameter as the soft measurement model was imputed to build T-S fuzzy model for measuring ball mill level. Firstly, fuzzy cluster and cloud model were used to identify premise structure and parameters respectively in order to conform fuzzy concept and rules; secondly, result parameters are identified by using a least square algorithm; Finally, fuzzy inference method was adopted for conducting the soft sensor of the ball mill level. The result of the experiment in small size ball mill proved the effectiveness of T-S fuzzy model for measuring ball mill level. Compared with the traditional method, this method has the feature of high accuracy and good stability. The paper includes the following aspects:1. The methods and tendency of testing mill load was summarized. The features of the method and the advantages of soft sensor were analyzed. It is concluded that adopting the soft sensor is an effective and prospective method for testing mill load.2. In the research, the feature of the ball mill’s motion, and the connection between the ball mill level and the operational mechanism was defined. And it is concluded that the signal of vibration has the feature of uncertainty and nonlinearity.3. For the feature of the ball mill, adopting T-S fuzzy model for building mill model was proposed. T-S fuzzy model could be used for identifying nonlinear system and strong robust. T-S fuzzy model has been used in many different fields conducting forecasting and soft measurement.4. Adopting subtractive clustering and improving C-means clustering and cloud model for building fuzzy model was introduced. And then the built fuzzy model was used for soft sensor of ball mill. 5. Through the contrastive analysis of the experiment result of traditional method and the T-S fuzzy model, it is concluded that T-S fuzzy model gets good effect, and has better accuracy and stability.In the paper, the T-S fuzzy model was built and soft sensor of ball mill level was conducted, based on collecting the data of the experiment in site and analyzing and processing the data. This testing method not only has better accuracy and stability, but also has extremely importance in improving the effectiveness and steady running of ball mill.