欢迎来到天天文库
浏览记录
ID:34144188
大小:1.89 MB
页数:5页
时间:2019-03-03
《基于混沌时间序列的瓦斯浓度预测研究》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、2442011,47(10)ComputerEngineeringandApplications计算机工程与应用基于混沌时间序列的瓦斯浓度预测研究1,22,32张宝燕,李茹,穆文瑜1,22,32ZHANGBaoyan,LIRu,MUWenyu1.晋中学院计算机科学与技术学院,山西晋中0306002.山西大学计算机与信息技术学院,太原0300063.山西大学计算智能与中文信息处理教育部重点实验室,太原0300061.CollegeofComputerScienceandTechnology,JinzhongUniversity,Jinzhon
2、g,Shanxi030600,China2.SchoolofComputer&InformationTechnology,ShanxiUniversity,Taiyuan030006,China3.KeyLabofComputationalIntelligenceandChineseInformationProcessingofMOE,ShanxiUniversity,Taiyuan030006,ChinaZHANGBaoyan,LIRu,MUWenyu.Studyongasconcentrationpredictionbasedoncha
3、otictimeseries.ComputerEngi-neeringandApplications,2011,47(10):244-248.Abstract:Inrecentyears,minedisastersareoccuringfrequently,especiallygasaccidentswhichareoftenreportedinnewspa-pers.Gasaccidentsareusuallyaccompaniedbyahighconcentrationofthegas,sogasconcentrationpredict
4、ingofthefuturemomentisaneffectivemeansofgasaccidentforecastingandhasveryimportantsignificancetoproductionsafetyincoalmines.ThispapersimplifiesC-Cmethodonchaostheory,andreconstructsphasespaceusingtimeseriesofgasconcentrationvaluesbymonitoringfromfivecoalmines.Thepaperchoose
5、sanappropriatevalueforthedelaytimeandtheembeddingdimensionaccordingtotherealvaluesofgasconcentration,andthenforecaststhegasconcentrationofthenextmomentusingweightone-ranklocal-regionmethod.Theexperimentalresultsshowtimeseriesofgasconcentrationhaveobviouslythechaoticpropert
6、y.Theresultsalsoshowmoderatecalculationandbetterforecastresultswhenthelengthoftimeseriesis500.Themeansquareerrorofforecastresultsof100abnormalgasconcentrationvaluesis0.122024,thusthemethodcanbeusedforthepredictionofgasaccidentsandprovidebasisfordecisionmakingincoalmines,fo
7、rexample,takingmeasuressuchasventilationininatimelymanner.Keywords:timeseriesofgasconcentration;chaos;reconstructphasespace摘要:近年来各种矿难频发,特别是瓦斯事故时见于报端。瓦斯事故通常伴随着较高的瓦斯浓度,因此,预测未来时刻的瓦斯浓度是预测瓦斯事故的有效手段,对煤矿的安全生产具有十分重要的意义。对混沌理论中的C-C方法进行简化,并用这种方法对5大煤矿的瓦斯浓度监测数据构成的时间序列进行相空间重构,依据数据的实际情况确
8、定其最佳时延和嵌入维,然后用加权一阶局域法对下一时刻的瓦斯浓度进行预测。实验结果表明瓦斯浓度时间序列具有明显的混沌特性,且当时间序列长度为500时,计算量适中且预测结果较优,对5
此文档下载收益归作者所有