Research on Multivariate and Multiscale Data Fusion Methods for Mine Gas Early Warning Applications

Author:Wang Bin Guo

Supervisor:du xue dong

Database:Doctor

Degree Year:2019

Download:1

Pages:112

Size:7087K

Keyword:

In the application of smart mines,in order to achieve accurate gas prediction and warning,it is necessary to extract a complete set of integrated data from multiple data sets.Therefore,researching multivariate and multiscale data fusion methods for smart mine applications can greatly improve the accuracy of smart mine data,and can realize multivariate and multiscale gas data prediction and early warning of mine,which has very important research significance and application value.Based on the background,significance and research status of smart mines,this paper summarizes the practical problems encountered in the application process of smart mine data fusion methods,and gives detailed research contents and research plans for the corresponding problems.The main work of this paper includes:(1)The rough set and neural network are used as the technical support,and the mine gas data is taken as the research object.Through the analysis of mine data flow analysis,data fusion,rough set and neural network theory,the data fusion framework for smart mine application is given.According to the characteristics of gas detection data,obtain gas data information,clarify the relationship between data features,perform rule reduction,extract corresponding fuzzy rules,generate gas prediction index according to its rules,and generate corresponding fusion results.it mainly provides complete fusion data and decision support for mine gas prediction and warning,and constructs a fuzzy neural network data fusion model that can be applied to gas emission prediction.(2)This paper studied the data fusion method in detail based on the decision-making analysis of mine gas early warning,then proposed data fusion method model and improved fuzzy neural network mine gas data fusion prediction model based on rough set and fuzzy neural network.It mainly includes the adaptive research,model design and algorithm research of fuzzy neural network and rough set integration method.(3)This paper presents a data fusion parameter vector selection and optimization method based on thought evolution algorithm.The initial weights and parameters of the fuzzy neural network model are optimized,and the data fusion experiments are carried out through seven variables and thirteen variables respectively.(4)This paper presents a data fusion parameter vector selection and optimization method based on wolf group algorithm.The convergence ability and global search performance of the learning algorithm in the data fusion model are improved,and the original data fusion and improved data fusion model is applied to the assessment and prediction of mine security situation.Theoretical analysis and simulation results show that the data fusion technology of rough set and fuzzy neural network proposed in this paper can be effectively applied to mine gas warning,It has achieved the paper research goal.