Research On Coke Quality Prediction And Evaluation Model

As one of the important solid fuels in industrial production,coke has a far-reaching impact on the economic beneficial result and development motivation of the industry,making coke quality a core research indicator in the process of coking production.In order to study coke quality more scientifically and comprehensively,and to further optimize the coking production process,it is necessary to realize on-line monitoring and evaluation of coke quality.At present,there is a lack of equipment that can effectively complete online monitoring of coke quality,and a complete and appropriate coke quality evaluation system is also indispensable.Therefore,in-depth research can be carried out from the establishment of coke quality prediction and evaluation models.Aiming at the process mechanism and each sub-process of the coke production process,the problem that it is arduous to monitor the coke quality online in the actual coke production process is analyzed,and we select representative coke quality index variables as output variables.The index variables that have a major impact on coke quality as an input variable,a model for predicting the quality of coke has established.While establishing this model,we select the BP neural network to achieve the prediction of coke quality,so as to achieve the purpose of deeply heightening the predictive capability and exactness of this model.To deal with the complicated problems resulted from the random generation of the initial weights and thresholds,mind evolutionary algorithm（MEA）is integrated into it.Combined with the formidable holistic optimization function of the mind evolutionary algorithm（MEA）,the initial weights and thresholds are optimized,and a coke quality prediction model based on MEA-BP is generated.Compared with the results of the primitive BP model,we can get better prediction results.In the study of the coke quality evaluation model,the process capability index is the core,and the coke quality index data is non-normal and multivariate.A comprehensive evaluation model for the multivariate process capability of the non-normal coke quality data is established,based on the coke quality prediction results.We can acquire the evaluation result,and divide the coke quality into unequal levels in accordance with the set criterion,and complete the comprehensive evaluation of the coke quality.After simulation analysis,the coke quality prediction model based on MEA-BP has improved predictive capability and exactness.At the same time,the comprehensive evaluation model of multivariate process capability can accurately and specifically realize the evaluation of coke quality,based on the non-normal coke quality data.Therefore,this research can be made available for the actual coke production process,in addition,it also supplies a scientific proof to process improvement.