摘要
大型煤粉锅炉安全稳定高效清洁燃烧是保障电站锅炉节能环保的关键因素。研究炉内过程的非线性机理,分析可表征炉内过程的非线性特征参数,掌握实际设备运行期间非线性特征参数的变化规律,不仅可为工程实践提供理论依据,还可以提高实际锅炉运行的效能和稳定性,具有深远的社会和经济效益。
本文将非线性理论与电站锅炉现场试验研究相结合,共开展了4个方面的研究:
1、基于统计理论,开展了对炉内压力纵向及横向数据的统计分析,确认炉内压力可作为炉内过程的参考变量;确认炉内压力序列分布基本不服从正态分布形式;炉膛压力序列存在明显的序列相关性,相关阶数最高约为13阶,大多数样本为非平稳序列,且序列中季节性表现不突出。这是研究炉内压力序列的分形及混沌特性的基础。
2、开展了实炉压力序列的分形及多重分形特性研究。确认多变量影响造成的不可预测的间断性跳跃,是炉内压力波动服从分形分布特征的根本性原因;炉内压力样本的方差无统计意义。确认炉内压力序列是一个局部稳定,整体波动的综合过程,属于非随机序列,并具有明确的持续性或反持续性特征,存在显著性变化和显著性持续影响两个标度时间。炉内压力序列的分形维可表征影响炉内运动复杂性的主要因素,Hurst指数可表征炉内运行期间的实际状况。综合利用分形维和Hurst指数,可对实炉运行过程中影响炉内工况变化的各因素加以区分,有利于实炉的故障诊断和运行优化。实际炉内压力序列具有多重分形特征,其局部奇异性的完整描述是对Hurst指数、分形维等整体特征标度指数的有益补充与完善,其由局域标度特征导出整体行为的思想可为描述炉内过程提供了新的方法。
3、开展了实炉压力序列的混沌时间序列分析及混沌特性研究。确认炉内压力信号为包含着随机信号的混沌信号,计算获得最优延时为8s、最小嵌入维数为8维,在重构炉内过程相空间基础上,计算得到某300MW机组满负荷正常运行状态下,炉内运动关联维为6.56,存在正的lyapunov指数0.0194,Kolmogorov熵为0.297bits/s。基于反演理论对炉内过程运动方程组进行了重建。反演方程组和实际炉内过程具有相似的非线性特性。分析方程组可知,相对单变量变化而言,不同影响因素间的非线性交互作用是影响炉内过程稳定的主要因素。
4、对某600MW超临界直流锅炉实际运行条件下的非线性规律开展研究,通过现场试验及理论计算总结可知:
(1)变负荷过程中,锅炉炉内过程非线性特征参数的变化规律为:锅炉冷态通风时炉内压力近似为随机运动。锅炉点火后炉内压力则转变为具有正或反持续性的非随机运动。锅炉升负荷过程中,原始压力序列Hurst指数规律性不明显;差分序列的Hurst指数显示,在较低负荷阶段,随负荷升高,炉内运动的反持续性具有增加趋势;在较高负荷阶段,反持续会保持在较高水平,但随锅炉负荷提高略有降低。
锅炉冷态通风的分形维数均高于锅炉点火后各工况而相应差分序列却呈相反规律。升负荷过程中,炉内压力序列的分形维呈上升趋势,但差分序列的分形维变化不大。
随锅炉负荷的提高,炉内过程的奇异性具有增加趋势。奇异性指数在冷态及低负荷阶段趋于大值,在高负荷阶段趋于小值。实际炉内过程的奇异性在整体上小值占优,炉内压力序列结构趋于粗糙。
升负荷过程各阶段炉内运动混沌特征量的计算表明,冷态K熵均大于点火后各工况,但关联维却最小。随负荷增加,关联维和K熵均具有增长趋势。由关联维可知,描述炉内运动需要的维数约为6维,即决定炉内运动的独立参变量的个数约为6个。
(2)炉内过程非线性特征参数与锅炉经济性指标间存在较为密切的关系。由相关性、拟合和预测分析可知,与锅炉效率相关性较高的非线性特征量为H、K和Δf;与飞灰含碳量相关性较高的非线性特征量为K、H、Hd;与炉渣含碳量相关性较高的非线性特征量为K、H、和Δα。利用BP神经网络,可较好的构建起非线性特征指数与锅炉经济性指标间的映射关系。
总之,通过本文的工作,可大致了解以炉内压力为参考变量的炉内过程中存在的非线性特性及其运行规律。结合锅炉运行经济性和稳定性研究,非线性理论在电站锅炉运行优化中将具有深广的应用价值。但是本文工作还只是非线性理论在电站锅炉炉内过程研究中的初步应用,本文所得结论的正确性均受限于所取样本的代表性,还需要后期更深入的理论和应用研究加以验证。
Keeping combustion in a large-scale pulverized coal boiler safety,stability,high efficient and clean is the key to energy conservation and reducing emissions.So there are very important to study the nonlinear mechanism of the furnace process,analyze non-linear parameters of the furnace process,and get the non-linear characteristic during the boiler running.These researches could provide theoretical bases for engineering practice and increase the boiler performance and stability, have far-reaching social and economic benefits.
Combined nonlinear theories and experimental investigation,the paper is proceeded in 4 aspects:
1.Based on statistical theories,lots of longitudinal and transverse furnace pressure series are analyzed.As study shows,furnace pressure as a reference variable can be used to study furnace process.The furnace pressure distribution doesn't obey normal distribution basically.Obvious serial correlations exist in furnace pressure sequences.The largest correlated order in sequences approximately is 13.Most of samples are non-stationary series.Seasonal performance doesn't highlight in sequences.These analyses are the foundation to investigate the fractal and chaos characteristics of furnace pressure sequences.
2.Researches on fractal and multi-fractal characteristics of the furnace pressure sequence are carried.Multi-variable induced unpredictable intermittent jumps are the essential reason why the furnace pressure fluctuation subject to the fractal distribution.Furnace pressure's variance doesn't have statistical significance.The furnace pressure sequence is a integrated process with locally stable and overall volatility.It belongs to non-random sequence,has the features of clear continuity or anti-persistent,exists two scale time of significant change and significant continuing impact. The main factors which influence on the complexity of furnace process can be characterized by the fractal dimension.The actual state of the furnace during operation can be characterized by Hurst index.Comprehensive utilization of fractal dimension and Hurst index,various factors which effect furnace operating conditions during actual furnace operation can be distinguished.It is helpful to the furnace fault diagnosis and operation optimization.Furnace pressure series have multi-fractal characteristics.The complete description of local singularity is a useful complement and improvement to Hurst index and fractal dimension.The overall behavior of the furnace can be derived from scale of its local characteristics,which also provide a new approach to describe the furnace process.
3.Researches on chaotic time series analysis and chaos characteristics of the furnace pressure sequence are carried.Furnace pressure signal belongs to chaotic signals with random component,and its optimal delay time is 8s,the minimum embedding dimension is 8.Based on reconstructing phase space of furnace process,the parameters when one 300MW unit normal operates at full capacity are got,where,correlation dimension is 6.56;positive Lyapunov index is 0.0194;Kolmogorov entropy is 0.297bits/s.Based on the inversion theory,the motion equations of the furnace process are reconstructed.Rebuild equations has similar non-linear characters with real furnace process.Analysis shows non-linear interactions between different factors is a major factor affected the furnace process's stability.
4.Researches on non-linear characters of a 600MW supercritical concurrent boiler are carried by field test and theoretical calculation.
(1) During the boiler load charging,variation law of nonlinear characteristic parameters of the boiler can be concluded as:
The furnace pressure of cold air is approximately random motion.After the boiler ignition furnace pressure changes into a positive or anti-persistence non-random movement.During the up-loading,theHurst index of original series don't have any rule distinctly,but the difference series Hurst index show the anti-persistence character of the furnace process is increasing during low load stage while the up-loading,and it will keep in a high level and lower slightly in high load stage.
The fractal dimension of cold air is higher than that after ignition.The difference series fractal dimension has the anti-rule.During the uploading,the fractal dimension has increase trend,but difference series fractal dimension doesn't change much.
With the increase in boiler load,the furnace process singularity increase too. Singularity exponents have a trend to big value in low load and a trend to small value in high load.Overall singularity of furnace process small value is dominant;the structure of furnace pressure series has a trend to coarse.
With the increase in boiler load,the furnace process has its chaos character. Entropy K of cold air is bigger than that after ignition,Correlation dimension is smallest contrarily.With uploading,Correlation dimension and entropy K both have an increasing trend.By Correlation dimension,the dimension to describe furnace motion is about 6,in a other word,the independency variable number is 6.
(2) Between the non-linear characteristic parameters of the furnace process and the boiler economic indicators exists closely relationship.By relevance, fitting and forecasting analysis,the non-linear characteristics highly associated with boiler efficiency are H,K andΔf;the non-linear characteristics highly associated with fly ash carbon content are K,H andΔα.By BP network, non-linear characteristics and boiler economic indicators could have a nice mapping.
In the paper,the non-linear characteristics and its running law is gotten by using the furnace pressure as variable.Combined with boiler economic and stable operation research,non-linear theory has huge applied value in boiler operation optimization.But the work in the paper is a preliminary application;the result in the paper is Iimited by the samples' representation.The in-depth theoretical and applied researches are needed to validate it.
引文
[1]樊泉桂.锅炉原理.北京:中国电力出版社,2004.
[2]世界资源报告,1988,中国环境科学出版社,北京,p142.
[3]邓元凯,张永和,李明,包国举等.具有高节能减排潜力的炉内非线性过程双尺度优化技术[J].华电技术,2008,30(5):64-69.
[4]徐旭常,施学贵,陈昌和,王云山等.煤粉火焰稳定原理--“三高区”原理的实验验证和数值模拟分析[J].锅炉技术,1994,1:2-7.
[5]秦裕琨,李争起,孙锐.风包粉煤粉燃烧原理及实验研究[J].中国电机工程学报,2000,20(5):59-62.
[6]傅维标,何裕昆,韩洪憔,周明德等.大速差射流型双一次风通道通用煤粉主燃烧器[J].工程热物理学报,1994,15(2):210-213.
[7]Glarborg P,Jensen A D,Johnsson J E.Fuel nitrogen conversion in solid fuel fired systems[J].Progress in Energy and Combustion Science,2003,29(2):89-113.
[8]Arenillas A,Backreedy R I,Jones J M,et al.Modelling of NO formation in the combustion of coal blends[J].Fuel,2002,81(5):627-636.
[9]Hill S C,Smoot L D.Modeling of nitrogen oxides formation and destruction in combustion systems[J].Progress in Energy and Combustion Science,2000,26(4):417-458.
[10]Zhong B J,Shi W W,Fu W B.Effects of fuel characteristics on the NO reduction during the reburning with coals[J].Fuel Processing Technology,2002,79(2):93-106.
[11]苟湘,周俊虎,周志军,杨卫娟等.低NO_x改造方案中煤粉再燃风喷口位置的选择[J].中国电机工程学报,2008,28(20):1-7.
[12]聂其红,孙绍增,吴少华,等.新型水平浓淡风低NOx煤粉燃烧器在贫煤锅炉的应用研究[J].中国电机工程学报,2002,22(7):155-159.
[13]秦明,吴少华,孙绍增,等.六角切圆燃烧褐煤煤粉锅炉低NOx燃烧技术研究[J].中国电机工程学报,2005,25(1):158-162.
[14]曹红加,唐必光,许洪波.煤粉浓淡燃烧方式对电站锅炉NOx排放影响的试验研究[J].中国电机工程学报,2004,24(8):229-232.
[15]Kurose R,Makino H,Suzuki A.Numerical analysis of pulverized coal combustion characteristics using advanced low-NOx burner[J].Fuel,2004,83(6):693-703.
[16]A.Vikhansky,E.Bar-ziv.Meaurements and numerical simulations for optimization of the combustion process in a utility boiler[J].J.Energy Res.,2004,28:391-401.
[17]曹庆喜,吴少华,刘辉.采用选择性非催化还原脱硝技术的600MW超超临界锅炉炉内过程的数值模拟.动力工程,2008,28(3):349-354.
[18]付忠广,王志鹏,史亮亮.燃煤锅炉微油点火燃烧器的数值计算分析.工程热物理学报,2008,29(4):609-612.
[19]岑可法,樊建人,池作和,沈珞婵.锅炉和热交换器的积灰、结渣、磨损和腐蚀的防止原理和计算.北京:科学出版社,1994.
[20]Wells J,Riley G,Williamson J.Interactions between coal-ash and burner quarls Part 2:resistance of different refractory materials to slag attack in a combustion test facility[J].Fuel,2003,82(15-17):1867-1873.
[21]周俊虎,赵晓辉,刘建忠,杨卫娟等.锅炉内卫燃带上高熔点灰渣沉积机理分析.中国电机工程学报,2008,28(14):20-26.
[22]王全钢,阎维平,朱予东.600MW燃煤锅炉受热面污染特性试验研究,中国电力,2007,40(1):52-55.
[23]M.Shimoda et al.Predication Method of Unburnt Carbon For Coal Fired Utility Boiler Using Image Processing Technique of Combustion Flame[J].IEEE Transactions on Energy Conversion,Vol.5.No.4,December 1990:640-645
[24]M.Hunt,S.Williams,K.Hickok,et al.An Intelligent Systems Approach for on Line Power Plant Efficiency,Emissions and Maintenance Diagnostics[J].The 1994 EPRI Heat Rate Conference,May 3-5,1994,Baltimore,MD.
[25]R.C.Booth and W.B.Roland.Neural network-based combustion optimization reduces NOx emissions while improving performace[J].Proceedings of the American Power Conference,Vol.60(2),60th Annual Meeting,1998,Chicago,USA.667-672.
[26]杨宏民,唐晓军,刘勇等.人工神经网络用于燃烧火焰状态识别的研究[J].工程热物理学报,1997,18(4):521-524.
[27]郑楚光,赵海波.实际锅炉中煤粉燃尽率与飞灰含碳量的预测[J]. 华中科技大学学报,2001,29(增1):14-17.
[28]赵利敏,初利辉,张世辉.电站燃煤锅炉燃烧预示BP网络的实现[J].电站系统工程,1998,14(3):32-35
[29]周昊,郑立刚,翁安心等.火电厂锅炉热效率优化调整软件的开发[J].中国电力,2004,37(2):29-31.
[30]林波.基于遗传算法的优化燃烧控制系统研究[D],南京:东南大学,2001
[31]漆小波,周松锐,黄卫星,祝京旭,石炎福.上行气固两相流充分发展段流动行为的相似特性,化工学报,2006,03.
[32]李晓祥,石炎福,顾丽莉,余华瑞.液固及气液固三相并流系统的动力学行为和混沌产生的研究,化学工程,2004,05.
[33]马少华,华莹,李少白.基于相空间重构的锅炉炉膛火焰信号分析[J].热能动力工程,2007,22(4):440-442.
[34]周臻,张韵辉.锅炉再热汽温时间序列的混沌分析[J].华中电力,2006.6:21-24.
[35]E.Piskova,L.M(o|¨)rl.Characterization of spouted bed regimes using pressure fluctuation signals.Chemical Engineering Science,2008,63(9):2307-2316.
[36]Srdjan Sasic,Bo Leckner,Filip Johnsson Characterization of fluid dynamics of fluidized beds by analysis of pressure fluctuations,Progress in Energy and Combustion Science,2007,33(5):453-496
[37]Hisashi Nakamura,Aiwu Fan,Hideaki Minamizono,Kaoru Maruta.Bifurcations of stretched premixed flame stabilized by a hot wall.Proceedings of the Combustion Institute[J],2009,32(1):1367-1374.
[38]Beck,J.M.and V.A.Volpert,Nonlinear dynamics in a simple model of solid flame microstructure[J].Physica D:Nonlinear Phenomena,2003.182(1-2):p.86-102.
[39]项静恬,史久恩.非线性系统中数据处理的统计方法(现代应用数学方法)北京:科学出版社,1997
[40]何灿芝,罗汉.应用统计学.长沙:湖南大学出版社,2004.
[41]王振龙.时间序列分析.北京:中国统计出版社,2000
[42]孙霞,吴自勤.分形原理及应用.合肥:中国科学技术大学出版社,2003
[43]Mandelbrot.分形对象--形、机遇和维数.北京:世界图书出版公司北京公司,1999
[44]郝柏林.分形与分维,科学杂志,vol.38,pp.9-17,1991.
[45]敖力布,林鸿溢.分形学导论.呼和浩特:内蒙古人民出版社,1996
[46]李翔,关勇,乔艳芬.基于分形理论的电力负荷持久性分析及预测[J].电网技术,2006.30(16):p.84-88
[47]A.Benassi,S.C.,J.Istas,Identifying the multifractional function of a Gaussian process.Statistics & Probability Letters[J],1998.39:p.337-345
[48]李彤,商朋见.多重分形在掌纹识别中的研究[J].物理学报,2007.58(8):p.4393-4400
[49]王信峰,曹占辉,李言俊.空间目标RCS序列的多重分形分析[J].计算机测量与控制,2008.16(12):p.1934-1937
[50]马丽萍,石炎福,黄卫星,余华瑞.循环流化床颗粒浓度波动信号多重分形测度分析[J].高校化学工程学报,2002.16(05):p.496-501
[51]Byczkowski,T.,Nolan,J.P.,and Rajput,B.,Approximation of Multidimensional Stable Densities[J].J.of Multiv.Anal,1993.46(1):p.13-31
[52]Julio C.Ramirez Pacheco,D.T.R.A Tool for Long-range Dependent Analysis via the R/S statistic[J].Proceeding of the 15th international conference on computing.2006
[53]朱兰瑾,林诚,气固喷动床压力波动的Hurst分析[J].福州大学学报(自然科学版),2003.31(2):p.247-252
[54]牛苗任,梁钦峰,于广峰,王辅臣.气流床气化炉压力波动信号的R/S分析[J].燃烧科学与技术,2005.11(5):p.458-464
[55]颜世森,郭庆华,梁钦锋,于广锁.撞击式火焰噪声信号的分形特性分析[J].动力工程,2008.28(04):p.555-558
[56]刘式达,刘式适.分形和分维引论.1993,气象出版社:北京
[57]Falconer,K.J.分形几何中的技巧.1999,东北大学出版社:沈阳
[58]罗世华,曾九孙.基于多分辨分析的高炉铁水含硅量波动多重分形辨识[J].物理学报,2009.58(01):p.150-158
[59]葛家怡,周鹏,赵欣,刘海婴.睡眠脑电信号的多重分形去势波动分析[J]. 天津大学学报,2008.41(10):p.1148-1151
[60]李国宾,段树林,于洪亮.,关德林.发动机振动信号特征参数的多重分形研究[J].内燃机学报,2008.26(01):p.87-91
[61]张磊,徐耀,蒋晓东,梁丽萍.多重分形谱研究溶胶-凝胶SiO 2疏水减反膜的激光预处理作用[J].物理学报,2007.56(07):p.3833-3839
[62]李晓光,徐德龙,范海宏.大颗粒流化床中颗粒反射光纤信号的多重分形特性[J].西安交通大学学报,2005.39(11):p.1259-1263
[63]邵元智,钟伟荣,任山,蔡志苏.纳米团聚生长的多重分形谱[J].物理学报,2005.54(07):p.3290-3297.
[64]李平,胡可乐,汪秉宏.多重分形谱在材料分析中的应用研究[J].南京航空航天大学学报,2004.36(01):p.77-81。
[65]彭志科,何永勇,卢青,褚福磊.小波多重分形及其在振动信号分析中应用的研究[J].机械工程学报,2002.38(08):p.59-63
[66]Berliner,L.M.,Statistics,Probability and Chaos[J].Statistical Science,1992.7(1):p.69-90
[67]Sangit Chatterjee,M.R.Y.,Chaos,Fractals and Statistics[J].Statistical Science,1992.7(1):p.49-68
[68]Hilborn,R.C.,ed.Chaos and Nonlinear Dynamics.1994,Oxford University Press:New York.
[69]Parlitz,U.,Nonlinear Time-series Analysis,in Nonlinear Modeling-Advanced Black-Box Techniques,J.V.J.A.K.Suykens,Editor.1998,Kluwer Academic Publishers.p.209-239.
[70]Reiss,H.D.,The Analysis of Chaotic Time Series,in Philosoghy in Physics.2001,Georgia Institute of Technology:Georgia.p.219
[71]Sauer T.,Y.Y.,M.Casdagli,Embedology[J].Stat.Phys,1991.65:p.579-616.
[72]Sauer T.,J.A.Y.,How many delay coordinates do you need?[J]Bifurcation and Chaos,1993.3:p.737-744
[73]吕金虎,陆君安,陈士华.混沌时间序列分析及其应用.2002,武汉大学出版社:武汉
[74]Cao,L.,Practical method for determining the minimum embedding dimension of a scalar time series.Physica D,1997.110:p.43-50
[75]陈国华,马军海,盛昭瀚.动力系统实测数据相空间重构的改进[J].东南大学学报(自然科学版),2000.30(1):p.16-21
[76]刘明言,胡宗定.气液固三相流化床流区及其过渡的混沌分析[J]_化学反应工程与工艺,2000.16(4):p.363-367
[77]盛昭瀚,马军海.管理科学:面对复杂性--混沌时序经济动力系统重构技术[J].管理科学学报,1998.1(1):p.31-41
[78]赵贵兵,陈纪忠,阳永荣,流化床压力脉动信号时间延迟相关性.化工学报,2002.53(12):p.1281-1284.
[79]Yi,Z.,Modelling Nonlinear Dynamics of Time Series with Applications,in Department of Electronic and Information Engineering[J].2006,Hong Kong Polytechnic University:Hong Kong.
[80]Kurths J.,H.H.,An attractor in solar time series[J].Physica D,1987.25:p.165-172
[81]Sato S.,M.S.,Y.Sawada,Practical methods of measuring the generalized dimension and largest Lyapunov exponent in high dimensional chaotic systems[J].Prog.Theor.Phys.,1987.77:p.1-5.
[82]Grassberger,P.,Proccaccia,I.,Characterization of strange attractors[J].Physical Review Letters A,1983.50:p.346-349
[83]Schouten,J.C.,Takens,F.,van den Bleek,C.M.,,Estimation of the dimension of a noisy attractor[J].Physical Review E,1994.50:p.1851-1861
[84]Schouten,J.C.,Takens,F.,van den Bleek,C.M.Maximum likelihood estimation of the entropy of an attractor[J].Physical Review E,1994.49:p.126-129
[85]赵贵兵,石炎福,段文峰.从混沌时间序列同时计算关联维和Kolmogorov 熵[J].计算物理,1999.16(3):p.309-315
[86]王丽婕,廖晓钟,高爽,冬雷.并网型大型风电场风力发电功率-时间序列的混沌属性分析[J].北京理工大学学报,2007.27(12):p.1077-1080.
[87]肖楠,金宁德,基于混沌吸引子形态特性的两相流流型分类方法研究[J].物理学报,2007.56(9):p.5149-5156.
[88]Schneider,T.M.,B.Eckhardt,and J.A.Yorke,Turbulence transition and the edge of chaos in pipe flow[J].Phys Rev Lett,2007.99(3):p.034502
[89]Rupert,C.P.and C.T.Miller,An Analysis of Polynomial Chaos Approximations for Modeling Single-Fluid-Phase Flow in Porous Medium Systems[J].J Comput Phys,2007.226(2):p.2175-2205
[90]J.van der Schaaf,J.R.v.O.,F.Tankens,J.C.Schouten,C.M.van den Bleek,Similarity between chaos analysis and frequency analysis of pressure fluctuations in fluidized beds[J].Chemical Engineering Science,2004(59):p.1829-1840
[91]Rosso,O.A.,et al.,Distinguishing noise from chaos[J].Phys Rev Lett,2007.99(15):p.1541-1543
[92]Erguler,K.and M.P.Stumpf,Statistical interpretation of the interplay between noise and chaos in the stochastic logistic map[J].Math Biosci,2008.216(1):p.90-99
[93]闫世杰,曹继伟,王长义,一维时间序列重构无刷直流电动机混沌吸引子特征[J].微特电机,2008(05):p.18-21
[94]Annunziato M,Abarbanel H D I.Nonlinear dynamics for classification of multiphase flow regimes[J].Proceedings of the Third ICSC Symposia on Intelligent Industrial Automation(ⅡA'99) and Soft Computing (SOCO'99).Rochester,USA:ICSC Academic Press,1999
[95]黄建平,衣育红.利用观测资料反演非线性动力模型[J].中国科学B辑,1991(3):p.331-336
[96]东方锅炉(集团)股份有限公司DG1900/25.4-Ⅱ4锅炉说明书
[97]沈小峰,胡岗,姜璐.耗散结构论.1987,人民出版社:上海
[98]焦李成.神经网络系统理论[M],西安:西安电子科技大学出版社,1994,21-100
[99]叶海文.循环流化床锅炉建模与控制的人工神经网络方法研究[D],北京:清华大学,1997.
[100]K.F.Reinschmidt.Neural Networks:Next Step for Simulation and Control[J].Journal of Power Engineering,1991,(2):41-45
[101]G.Irwin.Neural Network Modeling of A 200MW Boiler System[J].IEE Proceedings Control Theory Application,1995,142(6):529-536.
[102]颜廷虎,钟秉林,黄仁.Bp神经网络在旋转机械故障诊断中的应用[J].东南大学学报(自然科学版),1993.23(05):p.16-20
[103]何小荣,陈丙珍,胡山鹰,朱振伟.一种新的BP神经网络培训方法[J].化工学报,1994.45(05):p.573-579
[104]殷春根,骆仲泱,倪明江,岑可法.煤的工业分析至元素分析的BP神经网络预测模型[J].燃料化学学报,1999.27(05):p.408-414
[105]王妍芃,林宗虎.改进BP神经网络在流型判别中的应用[J].热能动力工程,2001.16(01):p.63-65
[106]周树孝,彭黔荣,周勇,石炎福等.散式流化和聚式流化的BP 神经网络识别[J].高校化学工程学报,2004.18(04):p.459-464
[107]张胜寒,范永哲,陈颖敏.基于BP神经网络的汽轮机转子钢热脆化性能的预测[J].中国电机工程学报,2005.25(23):p.110-113
[108]周坤,杨建国,赵虹.BP神经网络对煤着火特性的预测[J].热力发电,2005(11):p.21-25