Research And Implementation Of Control Parameters Optimization Of Non-analytical Model Based On Flotation Process

Mineral resources are very important for national production and life,but mineral resources in nature are often not directly used,and mineral processing is needed before the next use.However,mineral resources in nature are often not directly used,and they need to be processed for further use.Mineral flotation is the most common method in mineral processing.Mineral flotation is also known as foam flotation.The raw ore is broken and ground into ore particles.The hydrophobic-hydrophilic properties of ore particles are different due to the differences in physical and chemical properties of the surface in the flotation column.The specific reagents are added to the flotation column and the air is introduced into the flotation column,so that the smaller density ore particles are attached to the bubbles to form a foam layer,and the larger density ore particles sink to achieve the purpose of material separation and complete mineral processing.Mineral flotation is a complex industrial process with multi-input and multi-output,high coupling,nonlinear,long time delay and long process.There are many factors affecting the state of the system.The control parameters of the flotation process system are generally adjusted by the experience of workers.This adjustment method has no unified standard and is subjective,so that the state of the flotation system is changed due to the individual reasons of the field workers,which leads to the instability of the flotation system and affects the quality of the final concentrate of the flotation system.Based on the practical industrial application,this paper improves the stability of flotation system and the grade of concentrate,and the following work is done :(1)Based on the actual technological process of nickel ore flotation system,the industrial process and related parameters are analyzed,and the error analysis and processing of the original data collected by the sensor are carried out.(2)Because the mineral flotation process is extremely complex,there are many factors affecting the final concentrate grade of the system,and the input of the model is more when predicting the export grade of the flotation system.Therefore,the artificial neural network modeling method is used to predict the grade of the flotation system.Before modeling,the data are standardized and the principal component analysis data are reduced.Finally,the BP model and Elman model are used to establish the export grade prediction model of the nickel ore flotation system.The error between the predicted value and the actual value of the copper grade and the nickel grade of the nickel concentrate is analyzed,and the Elman model is determined as the basis for the follow-up work.(3)The control parameter optimization problem of flotation process is based on the established neural network model,which is a non-analytical model and cannot be optimized by the traditional derivation operation.In this paper,the particle swarm optimization algorithm without derivation of the objective function is selected to optimize the control parameters.Since the key parameters of inertia weight and learning factor are constant,the classical particle swarm constant,the global search ability of the algorithm is not enough,and it is easy to fall into local optimum.Therefore,this paper uses dynamic key parameters to improve the particle swarm algorithm.The final benchmark function test and the actual flotation control parameter optimization simulation prove the effectiveness of the improvement.(4)According to the needs of the project,an optimization simulation software system is developed.The data source of the whole project and the underlying sensor data acquisition are standardized by the data acquisition system in the middle layer,and then stored in the database.The upper optimization simulation software realizes the reading of the original data and the application of the optimization results by interacting with the database.The optimization simulation system is developed based on MATLAB platform.The database data is read according to time,and then the optimized model is determined by comparing the data with the modeling data,and the range of control parameters is set.Finally,the improved particle swarm algorithm is used to optimize the control parameters.According to the different sources of control parameters,the software has three simulation methods,optimization results simulation,input parameters simulation and console input simulation.According to the simulation results,the control parameters are saved to the database to guide the field production.