Model Simulation And Operating Performance Assessment For Coal Slurry Flotation Processes

The coal slurry flotation process is a complex continuous production process with long process flow.The rough production mode makes the production process unstable,resulting in the unsatisfactory washing effect of coal.To improve the automation and intelligence level of coal slurry flotation process,keep the coal slurry flotation process in the optimal operating performance as far as possible,and then improve the production efficiency and comprehensive economic benefits of coal enterprises.This thesis focuses on the mechanism modeling and operating performance assessment of coal slurry flotation,a whole process dynamic mathematical model based on material conservation and an operating performance assessment based on siamese long shortterm memory network for supervised slow feature analysis is proposed.The main research contents of the thesis are listed as follows:(1)Most of the existing models of coal slurry flotation process control mainly focus on the foam flotation,which is difficult to reflect the dynamic characteristics of the circuit of actual coal slurry flotation process.In this thesis,a dynamic mathematical model of coal slurry flotation for process control is established,and a whole process simulation platform of coal slurry flotation process is developed based on dynamic mathematical model.Firstly,the working characteristics of the key equipment in each link of coal slurry flotation are deeply analyzed,the dynamic mathematical models of the corresponding process are developed,and the modules are compiled and encapsulated in Simulink.Secondly,in order to simulate the complete washing process from raw material input to clean coal output in the actual coal preparation plant,each module is identified connected back and forth according to the process requirements,and a complete coal slurry flotation process model is constructed.Finally,the simulation experiment shows that the model can reflect the characteristics of variable coupling,slow time-varying and process time delay in the actual coal preparation process,which provides a simulation verification platform for the research of advanced control methods and optimization strategies of coal slurry flotation process.(2)Aiming at the problem that the traditional data-driven method can not establish a stable and reliable operating performance assessment model when facing the complex characteristics such as data nonlinearity and dynamic timing in the process of coal slurry flotation.In this thesis,an operating performance assessment based on siamese long short-term memory network for supervised slow feature analysis is proposed.This method makes full use of the structural advantage of siamese network to process two time samples at the same time,and extracts the potential dynamic characteristic information in the data of coal slurry flotation process.By introducing LSTM layer into the siamese network structure improve the operating performance assessment model the ability of nonlinear modeling and provides valuable process timing information.By introducing the slow feature constraint term and comprehensive economic index constraint term into the objective function,the model is forced to learn the slow feature information related to the comprehensive economic index.In addition,for the nonoptimal operating performance,a non-optimal factor identification method based on the contribution graph of siamese long short-term memory network is proposed.The cause variables leading to the non-optimal process are identified by calculating the contribution rate of process variables to the non-optimal performance.Finally,the effectiveness of the proposed method is verified on the whole process simulation platform of coal slurry flotation process.