Dynamic Models Of Dense Medium Coal Using Multi-parameter Sensitivity Analysis And Parameter Optimization

Dense medium coal preparation(DMCP)is currently the main processing method for coal washing in China.Its dynamic model is the foundation of characteristic analysis,optimization and control of DMCP.Therefore,an accurate dynamic model is the key to achieve intelligent coal preparation reduce the damage to the ecological environment,and ensure the economic benefit of the coal preparation plant.The model based on the conservation of mass is a dynamic model with simple model structure,convenient solution.And a description of the relationship between raw material input,output and loss by using the mass balance relationship of the various material in the slurry.Based on this model,the control system design of DMCP process can be realized.However,there are uncertain model parameters in the process of establishing model.At present,these uncertain parameters are mainly obtained by experience or experimentation.Therefore,the model is difficult to accurately describe the actual dynamic process.To solve the above problems,this paper combines the parameter sensitivity analysis with the intelligent optimization algorithm and reinforcement learning according to the characteristics of the dynamic model of the DMCP process.Driven by actual industrial data,optimized within the range of model parameters,and established a mathematical model of DMCP driven by mechanism knowledge and data mixed.The main contents are as follows:(1)Aiming at the problem that the large number of uncertain model parameters of the dynamic model of the DMCP process,which increase the complexity and calculation cost of the subsequent optimization algorithm,various sensitivity analysis methods are compared and analyzed and the selection system is given in this paper.By analyzing various sensitivity analysis methods and combining the characteristics of the dynamic model of dense medium coal preparation process,the sensitivity analysis method of Sobol' based on variance is used to analyze the model parameters,and the key parameters are screened out.Under the premise of ensuring the accuracy of the sensitivity results,the minimum number of samples required is determined through experiments,and the multi-parameter sensitivity analysis of the dynamic model of dense medium coal preparation process is completed,and the slurry separation ratio and the ash ratio constant in the overflow and underflow play a leading role in the model.(2)Aiming at the problem that the traditional prey-predator optimization algorithm tends to fall into the local optimum due to the constant step size,to improve the accuracy of model parameter optimization,a reinforcement prey-predator optimization algorithm is proposed.It uses the reinforcement learning idea to make search individuals use neural networks to map their own state information to the change of search step,update the reward mechanism according to the change of search step,and constantly update the network weight in the process of evolution iteration,realize the adaptive increase or decrease of search step.Finally,the optimization performance is improved and the parameter optimization of dense medium dynamic model is realized.The proposed algorithm can not only be used to estimate model parameters,but also has certain reference value for solving other optimization problems.(3)The validity and accuracy of the proposed method and the established model were verified through experiments.The results show that the RPPO algorithm can improve the search ability for the optimal key model parameters,making the model outputs closest to the actual measurement data.The root mean square error of the proposed model and the standard deviation of probability density are smaller than the other models,proving the effectiveness and accuracy of the proposed model.The thesis includes 33 figures,6 tables and 84 references.