Research On Online Optimazation Of Hematite Grinding Classification Process Based On Knowledge And Date

Grinding process is the most important part of the beneficiation process.China,s hematite ore reserves is large,but the grade is low,.the embedding size is fine,the composition is complex,frequent particle size distribution fluctuation and hardness change make the grinding process difficult to ensure the quality of final grinding products.The on-line optimization of the grinding and classification process can reduce the size fluctuation of grinding products,and improve the quality of ore processing.Therefore,it's become more and more important for mineral processing enterprises to optimize the process by grinding to improve the quality of grinding products in the fierce market competition conditions.The large inertia,strong coupling and parameter variation of the grinding process make it difficult to establish an accurate mechanism model for the optimization of grinding process.Therefore,this dissertation studies an on-line optimization method based on empirical knowledge,and applies the method of case-based reasoning technology to ouline optimization of grinding process.By studying the mechanism of grinding classification process,the influencing factors of grinding process optimization are analyzed,and the input and output variable attributes of online optimization model are determined.A case-based reasoning-based process optimization framework is proposed.A two-level case retrieval method based on grid clustering is proposed,which reduces the number of case traversal cases and provides time guarantee for online optimization.The k-means clustering method is applied to the case database maintenance strategy,and a case-base maintenance method based on clustering analysis and nearest neighbor algorithm is proposed to reduce the redundant cases and guarantee the optimization of real-time case database retrieval.Due to the frequent fluctuation of ore feed quality,the boundary condition of grindingprocess optimization is frequent,and it affects the accuracy of optimal setting value seriously.In order to improve the efficiency of optimization,a knowledge-based and data-driven on-line optimization compensation method is studied in this dissertation.The Bayesian network method is applied to ouline optimization compensation of grinding process.The compensation model of Bayesian network is established based on the experts,experience and experimental data,which can compensate the setting amount of ore feed,grinding concentration and overflow concentration.Finally,the whole process of on-line optimization and on-line optimization compensation of grinding grade are simulated.The results show that this method has the feasibility.