Data Processing Model of Coalmine Gas Early-Warning System

r.2007Journal of china Unof Mining TechnolVol 17Availableonlineatwww.sciencedirect.comSCIENCEdDIRECTSJ China Univ mining technol 2007, 17(1): 0020-0024Data Processing Model of CoalmineGas Early-Warning SystemQIAN Jian-Sheng, YIN Hong-sheng, LIU Xiu-rong, HUA Gang, XU Yong-gangSchool of Information and Electrical Engineering, China University of Mining Technology, Xuzhou, Jiangsu 221008, ChinaLIbrary of China University of mining& Technology, Xuzhou, Jiangsu 221008, ChinaAbstract: The data processing mode is vital to the performance of an entire coalmine gas early-warning system, especially in real-time performance. Our objective was to present the structural features of coalmine gas data, so that thedata could be processed at different priority levels in C language. Two different data processing models, one with priority and the other without priority, were built based on queuing theory. Their theoretical formulas were determined via aM/M/I model in order to calculate average occupation time of each measuring point in an early-warning program. Wevalidated the model with the gas early-warning system of the Huaibei Coalmine Group Corp. The results indicate thatthe average occupation time for gas data processing by using the queuing system model with priority is nearly 1/30 ofthat of the model without priorityKey words: gas early-warning: data processing; queuing theory; priority model; high efficiencyCLC number: TD 761 Introductionwhose research objects were static safety factorsDisasters caused by gas explosions with harmfulAt present, the collection, transmission and proceffects and high mortality rates are the main reason essing of gas data are mainly completed by differentfor coalmine accidents and its reputation for unsafe kinds of coalmine safety monitor and control systemsworking conditions. Once they occur, they will not in our country. These systems are private and notonly cause heavy loss of life and property but also generally available, Some large coal enterprises haveresult in social instability and serious international already embedded them into their Safety Productionramifications. For a long time, many scholars all Information Integrated System(SPIIS) based on aover the world have carried out a great deal of re- system of integrated techniques in order to share insearch work on coal and gas outburst prediction and formation in the entire coal enterprise through a localobtained some positive results. A literature area network 6-8. Compared with early-warningsearch found some traditional and modern presystems based on static safety factors found indiction methods. These traditional methods calculated literaturesit is obvious that coalmine gas earlycertain parameters by using static data from special warning systems based on dynamic gas data gener-mine instruments, which require 2-3 hours to activate ated by SPIIs are more scientific, pragmatic and inthe prediction operation, given present coal and gas real-time. Gas early-warning systems based on dyoutburst prevention rules. Modern methods analyze namic gas data could include at least one thousandthe static samples on the basis of traditional methods gas measuring points. This makes a search for similarby using chaos, fractal, mutation or fuzzy set theory, conditions almost infinitely long and might requireartificial neural networks etc, which lead to new analysis of time series data for each measuring pointmethods of prediction and have become the focus of Therefore it is necessary for the gas early-warningresearch. Some authors discussed coalmine safety system to process and analyze data more efficientlyarly-warning system from a management viewpoint, and accurately Iname time the data processing中国煤化工Received 06 July 2006: accepted 11 November 2006CNMHGProject 70533050 supported by the National Natural Science Foundation of ChinaCorrespondingauthorTel:+86-516-83885993;E-mailaddressyhs@hbcoal.comQIAN Jian-sheng et alData Processing Model of Coalmine Gas Early-Warning Systemmode should remain vital to the performance of the whose values are lower than the warning value pre-entire system, especially to real-time performance. scribed by coal mine safety regulations, while it mustBased on queuing theory, this paper presents an effi- pay attention to the gas measuring points whose valuecient gas data processing model with priority, which are near or higher than the warning value prescribedwas applied to the gas early-warning system of the by the regulations. Given the time series data of theHuaibei Coalmine group Corp and has run in a stable measuring points, the status of the point is identified,fashion for about two yearssuch as the state of security, the state of adjustment orthat of dangerously explosive conditions, which can2 Analysis and Presentation of Gas Data help in forecasting conditions of coal mine safety,helps to avoid dangerous conditions and to preventThe SPiis generates and stores real-time data con- gas disaster. In order to achieve these objectives andtinuously, which includes analog data for measuring given previous analyses, the information of the gasthe amount of gas, wind velocity, oxygen, carbon measuring point should include the mine # themonoxide,carbon dioxide, temperature, negative sub-stations #, the corresponding channel #pressure, etc. and digital data for indicating the op- physical position of the gas measuring point,erational states (on/off) of ventilator, local faconcentrations, the time of sampling and the datahoister,water pump, etc. The early-warning system processing priority level in the early-warning system,should not need to care for the gas measuring pointswhich is defined, in C language, as followsstruct gaschar MineO/ coal mine No. */int SubstationNo:/ sub-station No, =/nt channeINo/ channel No *char location[401; /*the physical position of gas measuring point*/float data_ sample: /gas concentrations value *float time_sample data sampling timeprioritypriority value: 0 stand for non-priority, I for priority/3 Data Processing Queuing Models of Gas and the queue of gas measuring points is formed atEarly-Warning Systemthe average arrival rate 2, which will receive serviceat the average processing rate u in the early warn-Although each data measuring point of the coal ing systemmine safety monitoring and controlling systems issampled in almost equal interval. In addition to gasService window Ameasuring points, there are other analog measuringIntcgrated:Methanesystem programpoints for monitoring wind velocity, oxygen, carbonprogrampoint queuemonoxide, carbon dioxide, temperature and negativepressure, as well as digital states for monitoring ven-Fig. 1 Information processing model without prioritytilators, local fans, hoisters, water pumps, etc. Befor gas measuring pointcause of this variety of measuring points the earlywarning system program processes the data of gas Let N(t)=(i, stands for the state of the system atmeasuring points which come from several coalmines. In the final analysis it is a random choice intomoment t, where i is the number of gas measuringwhich queue a gas measuring point will enter. The points that is being processed, i.e., the queue lengthqueue length can be regarded as infinite given theof system. It is easy to prove that N()/2) abackward and forward circulation of the measuringbirth and death process 20-22points. So for the sake of simplification, it would beLet p(i; t)=PIN(=() and let P(i)=lim p(i; t)pragmatic to analyse the system processing modeaccording to a m/M/1 queuing model, i. e the arrival i20. According to Fig. 2, the following balanceinterval and received service time of the gas measur-ing data points are all subject to a negative exponenequations can be listed when p==<1, we havetial distribution and the service window is 1. theAp(O)=p(1)queuing system, models, with or without priority,arediscussed below(+)p(1)=p(0)+p(2)3.1 Queuing system model without priority中国煤化工As shown in Fig. 1, the integrated system program(CNMHGgenerates the gas measuring point data continuouslyJournal of China University of Mining& TechnologyVoL17No. IFig. 2 State diagram of birth and death process for queuing system without priorityand >P()=l, thus, P(i=(1-p)p3.2 Queuing system model with prioritySo the average number of gas measuring poihe integrated system program wich continuouslythe queue isgenerates the gas measuring data points with the dataQ=∑i()=∑-p)p=Pstructure described above and the two queues are repectively formed at the average arrival rate m, nand the average number of gas measuring pointswhich will receive service at the average processingwaiting in the queue israte u,, u2 according to the service rules providedby the priority judgment module in the early warningsystem, as shown in Fig. 319. The priority judgmentmodule determines which gas measuring point shouldAccording to the Little Theorem, the average wait- be processed by the information processing moduleg time of a gas measuring point isThis follows the preemptive service rules to allow theTgas measuring point with high-priority interrupting1(1-p)(1-p)the service being received by the point with low-priand the average time for staying in the systemority and the points with the same priority receivingservice according to the FCFS (first come, firstserved )rules(1-p)Service windowqueue with priorityEarlystem programR Methane measuring pointqueue without priorityFig 3 Information processing model with priority for gas measuring pointsProvided that the gas measuring point with priorityis marked Cl and the gas measuring point without p=p,+p2 A As sIpriority is marked C2, and CI and C2 arrive respectively and independently at a Poisson flow with(+2)p(0,0)=H1P(1,0)+2P(O,1rameters A, n2 and their service time are subject to J(+22+H)p(,0)=HP(+1, 0)+Ap(i-1, othe negative exponential distribution with parame(+2+2)P(0.j=2P(0,j-1)+P(,j)+2P(O.j+1)ters lu H2, pis the system utilization rate, i.e., the (+n,+u)p(i,j)=hp(i-1,j)+n,p(i, j-1)+u, P(i+1,D)system busy rate. 2 is the overall average arrivalrate of data, u is the overall average processing ratWe can refer to references [20-22] for the processof data and there are=M+M2, p=p,+pof solving these equations, which calculated p(i, j)by seeking an inverse transform of the home funcpI(u,x)LetN()=(i,j), stands for the state of the systemmoment t. where the number of cl measuri(1-n1-P2)(1-z)o(z)points isi and the number of C2 is J. It is easyPuo(x)-1]{-o(x)k-(1-x)o(z)}prove that N(), t 0 is the birth and deathprocess 20-221Let P(i,; t)=PIN(t)=(i,)) and let p(i, j)+1+2(1-=)-21+1+2(1-2-441O(7)=lim p(i,; t)i,j20. According to Fig. 4, the fol中国煤化工lowing balance equations can be listed whenCNMHG differentialcoefficient of the nome iuhcuonl y(u, z)r.2007Journal of chinaof Mining TechnolVol 17+1,00,-1(a)state (i, 0)as center(b)state(0, j) as center(c)state(0,0)as center(d)state (i, j)as centerFig 4 State diagram of birth and death process for queuing system with priorityP(i,j=Therefore, the average stay time of a C2 measuringi!j! Ou'azLet pi. stand for the probability that there areCI measuring points in the system. Then P. standsTo, =2T+2-for the probability that there are J C2 measuringpoints in the system, whose probability home func1+12P1tions are respectively(u, 1),y(1, z)From(3), letz>1, according to Roberta rules weA1(1-p1)22(11(1-p)4 Results and discussion(,)1-P=0∑(-n)pThe following parameters are based on the fact thatSo, for P =(1-p)Pl, the result is the same as large or medium-sized state coal enterprise and thatgenerally there are more than ten coal minesthe M/m/l queuing system with only one type of client, indicating that the existence of C2 measuringthere are more than 100 measuring points in the coal-points have no effect on Cl. This is objectively con- mine safety monitor and control system, for whichsistent with existing conditionsthe typical scan period is 20 seconds. So, using theSimilarly, the average number of gas measuring models above, the gas early warning systems of thepoints and the average number of gas measuringHuaibei mipoints waiting in the queue are deduced as follows:more than 1 000 measuring gas points. In the meantime the network time delay and the integrated systemPW=Prprocessing time delay should be considered as wellpIWe have taken 1 000 gas measuring points as anAccording to the Little Theorem, the average wait- example and carried out the following theoreticalng time and the average stay time of a gas measuringanalyspoint in the system areFirst, by searching for similarity among all meas-uring points in an almost infinite time series in thequeuing system models without priority, we can iden1(1-p1)Twi1(1-P1tify whether the system is in a state of security,an, adjusting state or tends towards a dangerous andThen from(1), we haveexplosive state. Suppose A=30 points/second andformulas above, the average stay time of a measuring(1-n-P2)1-x)o(x)(2) point is 0.2 seconds, therefore, to process 1 000 pointdata requires about 200 secondno(x)-11[1-o()z-(-z)o(x)Secondly, in the queuing system models withoutthets that haCalculates the differential coefficient of z from value lower than that prescribed by coal mine safety(2)and let z=1. After obtaining P j the average regulations do not need excessive processing.Hownumber of C2 measuring points can be decided as ever, we should pay attention to gas measuring pointsfollowsthat are close to or over the value prescribed by theregulations. Stesip"1-P.-p2L 1-p).P1+-2B中国煤化工 d and u2=400CNMHGble statisticswe obtain, the diuraluI tI as uu, and that forrnal of China University of Mining TechnoC2 0.95 Still suppose / =30 points/second, then Corp and the results obtained from its mines indicate2, =1.5 points/second, 1,=28.5 points/second. Simi- that the average time for data processing using thislarly, according to the deductions above, the average queuing system model with priority is nearly 1/30stay time of a measuring point CI is 29.85 millisec- that of the model without priority in the coalmine gasonds; the average stay time of a measuring point C2 early-warning systemis 5.60 milliseconds. So to process 1 000 point data4)Theoretical results from the analysis were veri-requires about 6.813 secondsfied by the gas early-warning system of the Huaibei5 ConclusionsCoalmine Group Corp. It took an average of 10 seconds for the gas early-warning system to process the1)Our objective was to present the structural fea- entire data set of the gas measuring points by theture of coalmine gas data in a form that could be model with priority, while it took 400 seconds with-processed, at different priority levels, in C language. out priority. Especially when the average gas meas-2)Based on queuing theory, two data processing uring point arrival rate a is larger than its averagemodels were presented, one with priority and the processing rate u(i.e.p>1 ) it turns out the pracother without priority. The theoretical formulas that tical results are better than those of the theoreticalcould calculate average occupation time of each gas analysis. The advantage of the model with prioritymeasuring point in the early-warning program were that it is clearly better than the model without prioritydetermined from the mml modelThe reason is that there was an ideal p <1 in the3)The model used in the Huaibei Coalmine Group process of the theoretical analysisReferences[1] Guo H J. Discussion on coal mine accidents causes. Safery in Coal Mines, 2005, 36(11): 75-76.(In Chinese)[2] Zhao s Q. The main cause of accidents in coal mine. Mine safety, 2005, (9): 84-86[3] Liu H Y, Zheng P E Comparative study of accidents situation and analysis methods between china and U.S. A coal mininindustry. Industrial Engineering Journal, 2005, 8(1): 37-44. (In Chinese)[4] Zhang T G Compositive Prevention and Cure Techniques for Coal Mine Gas. Beijing: China Coal Industry Publishing House,2003: 1-2.(In Chinese)[5] Nie B S, He X Q, Wang E Y, et al. Present situation and progress trend of prediction technology of coal and gas outburstChina safety Science JournaL, 2003, 13(12): 1-4.(In Chinese)16] Qi WS, Ling B C, Cai S J. Developing trend and perspective in the research of predicting the coal and gas outburst chineSafety Science Journal, 2003, 13(12): 1-4.(In Chinese)[7 Zhong M H, Dong M s, Chen B Z Neural network expert system for risk prediction of coal seams outburst. Xiang Tan Min-ing Institute Journal, 1998, 13(1): 1-5(In Chinese)[8] Qin S Y. Fuzzy gathering classification similar analysis for the coal and gas outburst. The Chinese Journal of geologicalHazard and Control, 2003, 14(4): 58-61[9] Wang K, YuQX A computerized recognition and forecast system for the risk of coal and gas outbursts. China safety Science Journal, 1997, 7(2): 1-5.(In Chinese)[10] Wang K, YuQX, Peng Y Z Dynamic characteristics of gas emission during boring and its relation with gas outburst danger[Il Wang K, YuQx, Peng Y Z. Application of non-linear theory in study of coal and gas outburst. Journal of Liaoning Technical University(Natural Science ) 2000, 19(4): 348-352.(In Chinese)[12] Mu Q G. Study on coal mine accident defence management with systemic viewpoint. China Mining Magazine, 2005, 14(734-37. (In Chinese[13 Xiao Q X. Study on the early-warning safety management system for coal mine ventilation. Mining Safery Environmental Protection, 1999, 3: 4-7. (In Chinese)[14] Zhang M. Study on the early-Warning Management System for Coal Mine Security [Master dissertation]. Taiyuan: TaiyuanUniversity of Technology, 2004.(In Chinese[15 Sun L F. Study on the early-warning system for coal mine safety. Opencast Mining Technology, 2005(supp): 78-83.(InChinese)[16] Yin HS, Qian J S, Hua G, et al. Research on integration of supervisory control systems for coal enterprise in intranet environment. Computer Systems applications, 2005, (3): 12-15. (In Chinese)[17 Yin H S, Dai L J, Chen J B, et al. Research of integrated supervisory control system in modern enterprises based on WEB[18/ ri plication Research of Computers, 2003, 20(5): 144-146.(In Chinese)[18] Yin H S, Qian J S, Hua G, et al. Research and achievement of coal enterprise safety and production dispatch and commandsystem based on CSCW. Coal Science and Technology, 2006, 34(2): 52-56.(In Chinese[19 Zhong Q, Zhang Q s Evaluation of message operation modes on air-ground Vhf data link gateway system for CAAC. ActaAeronautica Et Astronautica Sinica, 2001, 22(1): 51-53. (In Chinese)[20] Hua X. Queuing Theory and Random Serving System. Shanghai: Shanghai Translation Publishing Corporation, 1987.(InChinese)21] XuG H Random Serving System(Second Edition). Beijing: Science Publishing中国煤化工[22] Tang Y H, Tang X W. Queuing Elements and Applications. Chengdu: UniverChina Publishing House, 2000.(In ChineseCNMH