Multi-granularity Representation Method Of Big Data In Coal Mine Safety Based On Cloud Model And Its Application

Since coal mine safety accidents happen frequently,how to mine useful knowledge from massive coal mine safety data is particularly important.At present,the processing of coal mine safety production data is generally based on the quantitative model,which has a single solution idea and can not meet the multi-granularity control mode under the coal mine safety supervision framework,so it is impossible to objectively and comprehensively mine the risk information contained in the coal mine data.How to recognize and predict its risk situation from the perspective of multi-granularity under the massive data resources is an urgent problem to be solved at present.This article is based on cloud model to have a transformation between the quantitative and qualitative data of coal mine safety production,combined with the corresponding coal mine production safety architecture,from the time dimension,spatial dimension to mine risk knowledge.And it is applied to coal mine gas risk prediction.The main research work is as follows:1.Spatiotemporal multi-granularity representation method for coal mine safety big data.Based on the variable granularity control mode of coal mine safety production,the production data at different granularity levels are granulated,and further combined with the actual production mining area to represent and analyze with different cognitive perspectives,which lays a foundation for the subsequent analysis,interpretation and modeling of coal mine data.Based on the above actual production requirements,take advantage of the cloud model,from the time dimension,spatial dimension and combined with focus in the field of production to give a formalize method for different grain layer,further giving the risk of semantic cognition as a result for time grain and space grain,and realize the space-time multi-granularity representation and cognition of coal mine safety big data.2.Multi-granularity concept extraction method of coal mine safety Big data.Consult the classic concept extraction method-adaptive gaussian cloud transform(A-GCT),found that skewness distribution in coal mine serious data sets is easy to appear the concept extraction is not complete,especially in the coal mine security situation over this part of the early warning value data,gaussian cloud transform often cannot extract the concept of this kind of high risk,and in theory domain boundaries,its border to extract concepts will membership cognition unclear problems.Aiming at these two questions,to have a logarithmic transformationthe for thr original distribution function firstly,and invoke the gaussian mixture model to convert frequency distribution into a superposition of multiple gaussian distribution,and join in the generation of normal distribution interval optimization,the concept of form is further concluded according to the ambiguous relationship between the distribution.In addition,the generated concept of the domain boundary was fitted into a semi-trapezoidal cloud to further optimize the unreasonable cognition of the boundary concept in the original method,and an adaptive hybrid cloud transformation algorithm was proposed to realize the multi-granularity concept extraction of coal mine safety big data.3.Multi-granularity coal mine risk conceptual prediction based on time-varying cloud.The current forecasting models are generally based on quantitative forecasting,which makes the prediction often deviates greatly under the influence of external factors,and the existing qualitative prediction method in the treatment of coal mine data on the random oscillation of often accuracy is not high,the error of the ultimate assessment also adopt quantitative evaluation index,which to some extent ignoring the qualitative representation of the fuzziness and the integrity of the concept and aiming at these problems,multiple granularity time varying cloud model is put forward,in the timevarying cloud model of existing before joining the concept extraction and multigranularity said,In order to improve the prediction accuracy of random oscillation series such as coal mine data,the smoothing operator is added,and the concept similarity measure and parameter error are put forward as the prediction indexes,which makes up the disadvantage of using only parameter error as the prediction index,and be applied in the coal mine gas risk warning.