Research On Operation Optimal Control Of Grinding Process

In the metallurgical and mineral processing industry,the grinding process is an important process after the crushing process.Its function is to continue grinding the crushed large particle minerals to the appropriate particle size,so as to dissociate the useful minerals from gangue monomers or different useful minerals from each other,so as to provide qualified particle size slurry for subsequent separation.The grinding process mechanism is complex,with comprehensive complex characteristics such as nonlinearity,slow time-varying,large lag,and the key process parameters cannot be detected online and in real time,which greatly increases the difficulty of modeling and calculating the set value of operation quantity.Under the trend of increasing requirements for product quality and saving production cost in the grinding process,the traditional detection methods and optimization methods have been unable to meet the current production needs.The unreasonable problems of real-time and accurate detection and set value of secondary grinding particle size have aroused extensive discussion in the beneficiation industry.To solve the above problems,the work of this thesis is as follows:(1)By reading a large number of literature and books related to the production of ball mill,we have a certain understanding of the grinding process of ball mill and the classification principle of hydrocyclone.The mechanism model of ball mill grinding process and hydrocyclone classification process is understood.Through going to the production site to communicate with operators and experts,combined with the knowledge of mechanism model,many factors affecting the particle size of secondary grinding are analyzed,and the input required for modeling is selected.(2)Based on the actual production data in the concentrator,the two-stage grinding particle size soft measurement model is established.There are many interference factors in the grinding process,resulting in the uncertainty between the variables of the two-stage grinding particle size model.T-S fuzzy neural network combines the characteristics of fuzzy system and neural network,and can deal with the uncertainty rules in the data collected by the factory.For T-S fuzzy neural network,the random initialization of network parameters is often used,and the value of network parameters will affect the accuracy of the model.In this thesis,PSO is used to optimize the weight of T-S fuzzy neural network,improve the learning ability of T-S fuzzy neural network,and then improve the accuracy of the model.RBF neural network,T-S fuzzy neural network and PSO optimized T-S fuzzy neural network are used to model the two-stage grinding particle size respectively.From the comparison results of the three models,it can be seen that the prediction effect of PSO optimized T-S fuzzy neural network model is the best.(3)In view of the problem that the set values of two important operating variables,ore feed and water feed at the mill inlet,are still given artificially,this thesis uses genetic algorithm and PSO optimization T-S fuzzy neural network model to optimize the set values.The optimization goal is to maximize the proportion of the qualified particle size of the second stage grinding,and the 1# ball mill current is taken as the constraint condition to obtain a more reasonable set value.It can be seen from the results that the qualified rate of the optimized two-stage grinding particle size products is significantly increased,and the selection of appropriate ore feeding amount and water feeding amount improves the processing capacity of the ball mill.In the case of safe production of ball mill,it not only ensures the output and quality,but also improves the production efficiency,which makes preparations for the subsequent application of excellent automatic control strategy in the grinding production process.Taking the grinding production equipment in the grinding process of hematite in a concentrator as the research object,this thesis puts forward a model that can detect the two-stage grinding particle size in real time,takes the maximum qualified rate of the two-stage grinding particle size as the optimization goal,and uses genetic algorithm to determine the best set value of ore feeding and water feeding of operating variables,so as to achieve the purpose of safe production,reducing energy consumption and saving cost,It provides a new idea in the field of operation optimization control of hematite grinding process and particle size modeling of two-stage grinding.