| Grinding and classification process is one of the important links in the process of mine-selecting, and the quality of grinding classification work directly affects the economic benefits of the concentrator. However, the characteristics of multivariable, nonlinear and uncertainty (such as the properties of ore, the structure of mills and the operation of grinding) etc lead to the results of unstable grinding concentration, the low efficiency of ball mill production and hydrocyclones, the non-compliance of grading granularity and concentration, and so on. Therefore, in recent years, how to make accurate analysis and good control on the grinding classification process becomes the focus of attention.The traditional control method in the process of grinding classification is to control the feeding capacity, the grinding concentration and the particle size of overflow with the weakness of unsatisfactory of control on system performance index, such as the ball mills cannot work fully, the grinding concentration is unstable, the efficiency of classification is low, and so on. As a result, the present paper proposes a control method basing on the fuzzy neural network. On the basis of the studying and designing of optimal feeding capacity to mills, pulp density to sand pump pools and the pressure of cyclone inlets, this method select the particle size of overflow, the pressure of cyclone inlets, the feeding capacity, the grinding concentration of sand pump pools, the mill power and electric ear as the tested items on one hand, and select the pressure of cyclone inlets, the latter water addition and the feeding capacity as the manipulated variable on the other hand, in order to achieve the automatic control of the feeding capacity on ball mills, the control of the grinding concentration and the particle size of overflow of cyclone inlets.Basing on the technology and the control mode of grinding classification, the present paper mainly makes an analysis on the overall scheme, equipment selection, hardware design, software design and the control algorithm of the control system of grinding classification. The hardware system involves industrial computer, programmable controller and its extension modules, testing equipments such as electromagnetic flow meter, densitometer, level gauge, PSI200particle size analyzer, etc., transmission devices such as solenoid valves, electronic scale, frequency converter, etc. Finally, implementation steps and program code of the fuzzy neural network on-line control in the process of grinding classification are introduced; the comparison of the conventional control method and the fuzzy neural network method in controlling the feeding capacity and the particle size of overflow is showed; as well as the advantages and disadvantages of the control is summarized. Fameview configuration software is adopted by the system software design for completing system monitoring and management of picture as well as the OPC communication method between configuration software and Matlab is adopted for the fuzzy neural network algorithm by transmitting the PLC collected data in Matlab. Then, the computing results in Matlab would be back to the register of PLC and the system would be controlled by PLC. This way can give full play to the PLC and Matlab’s advantages and make the advanced control algorithm not only limited to the off-line simulation.The control system in this paper is being studied and designed under the background of the grinding classification system of HUIZE dressing plant which is one of the subsidiary of YUNNAN CHIHONG Zn&Ge CO., LTD. And by this way, the control method basing on the fuzzy neural network has been applied successfully in the grinding classification system by reaching the purposes of increasing the machine handing capacity of ball mills, stabilizing the grinding concentration, improving the quality of grinding classification products, reducing the energy consumption and production cost, and has a certain practical application value. |
PDF Full Text Request