| China is rich in coal resources,accounting for 13 percent of the world’s total coal.Coal has been one of the main energy sources in China for a long time.This traditional energy structure model dominated by coal will not change in the short term,and it is estimated that coal will account for more than 50% of China’s primary energy production and consumption by 2050.The safety of coal mining concerns people’s livelihood and has always been a highly valued issue in China.However,the hydrogeological condition of China’s coal mines is complex,and with the increasing intensity and depth of coal mining,the disastrous problems of coal geology seriously threaten the safety of production.Among the five mining accidents of gas,water damage,roof,fire and dust,mine water inrush is second only to gas accident,with relatively high occurrence frequency and serious hazard.Rapid and accurate identification of water inrush sources is the key to prevent water inrush accident.The most commonly used method to identify water source of water inrush in coal mine is routine hydrochemical identification.The development of computer technology and mathematical geology provides a means to eliminate interference and quantify the inrush water source identification.In recent years,there are many computer methods and researches on water inrush source identification based on the analysis and research of conventional water chemical indexes.However,these methods have the following problems in the actual rapid identification:(1)Poor applicability of evaluation index.The distinguishing index of mine water source type generally uses eight conventional ions of groundwater,ignoring the difference of mining areas and mines in the same mine area.The differences of hydrogeological conditions and their increasingly complicated status are also ignored.(2)The discretization process reduces speed and accuracy.At present,the attribute reduction algorithm applied to mine water source identification is only applicable to the processing of discrete data.When processing continuous data,continuous data needs to be discretized,which inevitably leads to loss of information,reduced accuracy and speed of recognition.(3)Big Data processing ability is weak.With the increase of distributed computing ability,real-time monitoring and recognition process will face big data processing,and in the face of high-dimensional data,so-called dimension disaster is often caused.(4)Universality research needs to be deepened.(5)Security needs to be improved.Data transmission of mine water source is required between local coal mines and relevant mining bureaus and other units.Online identification and early warning system need to transmit massive data in real time,which will inevitably lead to information security problems in the process of data reading and transmission.In view of the above situation,taking the water source characteristics of different aquifers and the identification algorithm of mine water inrush in Chensilou coal mine as the research object,two methods of rapid identification of mine water source are proposed.The main research results and conclusions are as follows:1.On the basis of in-depth study of normalized cheeger cut,graph P-laplacian,neighborhood rough set,attribute reduction and spectral clustering theory,normalized cheeger cut based on neighborhood rough approximation(NRA-NCC)is proposed for rapid and accurate identification of mine water inrush.This algorithm is suitable for analyzing and processing massive high-dimensional mine water inrush data.It solves the problems of poor applicability,dimension and precision reduction,and weak ability of large data processing by four-way processing.They are as follows:(1)Increasing the index of water resource to improve the comprehensiveness and applicability.(2)Attribute reduction based on domain rough set can avoid the discretization of continuous data and improve the speed of identifying water source types.(3)Attribute reduction introduces information entropy to improve the accuracy of water inrush source identification.(4)The NRA-NCC algorithm for clustering is proposed to deal with the real-time acquisition of high-dimensional water source data in mines.2.This paper proposed a new method based on feature importance to identify the mine water inrush source in random forest rapidly which is suitable for analyzing and processing a small amount of mine water inrush data with missing items.The traditional recognition algorithm based on machine learning is suitable for processing and analyzing a large amount of data.However,it is difficult to obtain a large amount of training sample data in the mine water inrush problem.Combining with the actual situation of Chensilou Mine,this paper presents a rapid Identification of Water Inrush Source in Random Forests Based on Feature Selection.When the training samples are scarce,the algorithm can still identify the outstanding water resources quickly and accurately,and can maintain a high accuracy,and it is not easy to produce over-fitting.3.The increasingly prominent problem of mine water damage puts forward higher technical requirements for rapid identification of mine water inrush sources.The twelfth five-year plan national science and technology support project supported by this study(real-time monitoring and early warning system for major mine water inrush disasters)is a new method and theory for information processing of mine water inrush source based on the Internet.Information security is involved in real-time identification,remote analysis,cloud computing platform processing and distributed computing of mine water inrush source data,and there are relatively few researches in this field at present.This paper proposes an anonymous authentication scheme using zero-knowledge proof of digital signature based on Wang et al.The security and computational efficiency of this protocol are verified by experiments.The scheme not only reduces the communication flow of mine water inrush prediction and early warning system data transmission,but also has a higher level of security. |
PDF Full Text Request