Study On The Intelligent Control Of Flotation Process Based On Deep Learning

The flotation process is a non-linear system and the relationship between various factors can not be determined by a simple model.At the same time,due to the lack of accurate detection devices,some factors that affect the flotation process can not be included in the model study,resulting in the poor accuracy of the slime flotation prediction model for a variety of reasons,thus affecting the accuracy and stability of the flotation process control.Therefore,this paper collects the actual data of the flotation field,tries to establish a more accurate prediction model of the flotation process by selecting the appropriate algorithm,designs the control strategy more in line with the flotation process,and studies the intelligent control of the flotation process.In this paper,the flotation process of ×× coal preparation plant is taken as the research object,and the parameter collection platform of flotation process was constructed;the collected data were preprocessed,lagged and analyzed to obtain the flotation process parameter data that can be used for model prediction;based on the deep learning framework,DNN and GRU algorithm were used respectively to establish dosage prediction model and ash prediction model of flotation process.Through error analysis,the GRU prediction models were selected as the flotation process prediction model.The iterative learning control was introduced to build the flotation process intelligent control strategy,and the flotation process intelligent control system based on the depth learning was designed.Combined with the original detection instrument and control network in the coal preparation plant,a parameter collection platform for the flotation process was constructed,the data sources and data collection and storage methods were defined,and the real-time and historical data collection of the flotation process parameters were realized.Finally,seven sets of parameter data,including coal type,raw coal quantity,raw coal ash content,foaming dose,collection dose,floating fine ash content and tailings ash content,are selected for research.Pure lag processing,data preprocessing and correlation analysis were carried out for the collected data.Data preprocessing mainly included outlier cleaning and missing value filling.Correlation analysis clarified the input and output variables of flotation prediction model,and determined the model structure as "5 input,2 output".Based on the DNN algorithm,the flotation dosage prediction model and ash prediction model were established.The error analysis of the model test results was carried out to determine the cause of the error and optimize it.In view of the lack of data,the paper introduced the time factor,used GRU algorithm to build the flotation prediction model,make a comparative analysis of the two groups of models,determined that the prediction quality of the model is optimized,and finally get the neural network prediction model of flotation process based on deep learning.Based on the original PID feedback control,the iterative learning control method based on neural network model prediction was introduced to construct the intelligent control strategy of flotation process.Finally,on the basis of the previous research,an intelligent control system for flotation process is designed,which includes four parts: data acquisition and storage,data processing and analysis,prediction model simulation and control module.There are 34 figures,32 tables and 67 references included in this paper.