转炉炼钢过程脱磷和吹氧模型的研究
|
摘要
脱磷和吹氧过程是转炉炼钢的两个重要环节。磷是有害元素,对钢材的性能有许多不利影响,因此必须使磷脱除到一定水平以下。吹氧则是转炉炼钢的主要控制方式,通过向转炉中吹入氧气,将铁水中的碳、硅、磷、硫等元素脱除,并利用氧化反应放出的热量,使熔池的温度升高,进而得到合格的钢水,因此合理地控制吹氧量是非常必要的。但由于转炉炼钢具有炉内反应复杂,生产条件多变、原料不稳定、生产数据的整体规律性不强等特点,给吹氧量和熔池中磷含量的控制带来了很大困难。本文采用模糊c均值、粗糙集、案例推理等算法建立了脱磷模型和吹氧模型。
脱磷模型方面,本文通过对转炉中脱磷反应的热力学分析,推导出钢渣与钢水之间磷分配比的基本关系式,并提出对多个模型模糊加权的方法来计算磷分配比。该方法首先采用模糊c均值聚类算法将炉况相近的数据进行聚类,然后分别对各分类结果建立多元回归磷分配比模型,并计算出数据样本对应各模型的隶属度,作为模糊权值,最后将各磷分配比模型的计算结果模糊加权估计钢渣与钢水间的磷分配比。采用实际生产数据进行仿真,结果证明该脱磷模型的有效性。
吹氧模型方面,本文采用案例推理的方法建立吹氧量计算模型,主要包括案例检索和案例调整两个过程。在案例检索过程中,针对传统贝叶斯粗糙集理论只能处理二决策类的不足,提出了一种基于多决策类的贝叶斯粗糙集,在此基础上定义了一个衡量条件属性对决策属性影响程度的γ依赖度函数,并证明该函数具有随着条件属性的增加而单调递增的性质,进而基于γ依赖度函数的单调特性提出了一个确定属性权重的算法,用以确定吹氧量各影响因素的权重,在此基础上利用最邻近法检索出与当前炉次情况相似的一组案例做为参考。在案例的调整过程中,采用检索出的相似案例训练分层混合专家模型,并用微粒群算法优化模型参数,将当前炉次的情况代入到这个专家模型,进而计算出当前炉次的吹氧量。采用实际生产数据进行仿真,结果证明该吹氧模型的有效性。
The processes of dephosphorization and oxygen blowing are two important parts of steelmaking in basic oxygen furnace. Phosphorus is a harmful element, which has many adverse effects on the performance of steel. Therefore, it is necessary to make the phosphorus content below a certain level. Oxygen blowing is the main control method in steelmaking, oxygen is blowed into the converter, then carbon, silicon, phosphorus, sulfur and other elements in the molten iron are removed by oxidation reaction, and the heat of oxidation is released, so that the bath temperature increases, at last qualified steel is got, so reasonable control of blowing oxygen is necessary. However, the steelmaking has many characteristics, such as complicated reactions, ever-changing production conditions, instability raw material, the overall regularity of production data is not strong, which brought great difficulties in controling oxygen volume blowing and phosphorus content. This paper make use of fuzzy c-means, rough sets, case-based reasoning algorithms to establish the dephosphorization model and oxygen blowing model.
In the aspect of dephosphorization model, Based on the thermodynamics analysis of dephosphorization process in converter, this paper gives basic relationship of slag-metal phosphorous partition ratio, and proposes a method of multiple models by fuzzy weighted to calculate slag-metal phosphorous partition ratio. First, fuzzy c-means algorithm is used for clustering of data which is in the similar condition of the furnace, and get multiple regression phosphorous partition ratio models for each classified result. Then calculate the degrees of membership of the data to each model as fuzzy weights. Last, the fuzzy weighted method is used for the calculating results of each phosphorous partition ratio model to estimate the slag-metal phosphorous partition ratio. The simulation result of the model using the practical data from a steel plant proves the effectiveness of this dephosphorization model.
In the aspect of oxygen blowing model, this paper uses case-based reasoning approach to calculate oxygen volume blowing, which mainly contains the processes of case retrieval and case adjust. In the case retrieval process, focusing on the limitation that traditional bayesian rough set model theory can only deal with the situation of two decision classes, a improved bayesian rough set model is proposed to deal with the problem of multiple decision classes. On this condition, a y dependency function is defined to evaluate the condition attributes significance to decision attributes, and it is proved that the y function is monotonic increase with condition attributes. In the end, an algorithm to compute attribute weight is proposed based on the monotonic property of y dependency function, using which to determine the weights of factors affecting oxygen blowing. On this basis, the most nearest neighbor method is used to retrieve a group of similar cases as the current furnace occasion. In the case adjust process, it trains the hierarchical mixture of experts model with this group of similar cases, then particle swarm optimization algorithm is applyed to optimize the model parameters. Finally, calculate oxygen volume blowing in current production condition. The simulation result of the model using the practical data from a steel plant proves the effectiveness of this oxygen blowing model.
脱磷模型方面,本文通过对转炉中脱磷反应的热力学分析,推导出钢渣与钢水之间磷分配比的基本关系式,并提出对多个模型模糊加权的方法来计算磷分配比。该方法首先采用模糊c均值聚类算法将炉况相近的数据进行聚类,然后分别对各分类结果建立多元回归磷分配比模型,并计算出数据样本对应各模型的隶属度,作为模糊权值,最后将各磷分配比模型的计算结果模糊加权估计钢渣与钢水间的磷分配比。采用实际生产数据进行仿真,结果证明该脱磷模型的有效性。
吹氧模型方面,本文采用案例推理的方法建立吹氧量计算模型,主要包括案例检索和案例调整两个过程。在案例检索过程中,针对传统贝叶斯粗糙集理论只能处理二决策类的不足,提出了一种基于多决策类的贝叶斯粗糙集,在此基础上定义了一个衡量条件属性对决策属性影响程度的γ依赖度函数,并证明该函数具有随着条件属性的增加而单调递增的性质,进而基于γ依赖度函数的单调特性提出了一个确定属性权重的算法,用以确定吹氧量各影响因素的权重,在此基础上利用最邻近法检索出与当前炉次情况相似的一组案例做为参考。在案例的调整过程中,采用检索出的相似案例训练分层混合专家模型,并用微粒群算法优化模型参数,将当前炉次的情况代入到这个专家模型,进而计算出当前炉次的吹氧量。采用实际生产数据进行仿真,结果证明该吹氧模型的有效性。
The processes of dephosphorization and oxygen blowing are two important parts of steelmaking in basic oxygen furnace. Phosphorus is a harmful element, which has many adverse effects on the performance of steel. Therefore, it is necessary to make the phosphorus content below a certain level. Oxygen blowing is the main control method in steelmaking, oxygen is blowed into the converter, then carbon, silicon, phosphorus, sulfur and other elements in the molten iron are removed by oxidation reaction, and the heat of oxidation is released, so that the bath temperature increases, at last qualified steel is got, so reasonable control of blowing oxygen is necessary. However, the steelmaking has many characteristics, such as complicated reactions, ever-changing production conditions, instability raw material, the overall regularity of production data is not strong, which brought great difficulties in controling oxygen volume blowing and phosphorus content. This paper make use of fuzzy c-means, rough sets, case-based reasoning algorithms to establish the dephosphorization model and oxygen blowing model.
In the aspect of dephosphorization model, Based on the thermodynamics analysis of dephosphorization process in converter, this paper gives basic relationship of slag-metal phosphorous partition ratio, and proposes a method of multiple models by fuzzy weighted to calculate slag-metal phosphorous partition ratio. First, fuzzy c-means algorithm is used for clustering of data which is in the similar condition of the furnace, and get multiple regression phosphorous partition ratio models for each classified result. Then calculate the degrees of membership of the data to each model as fuzzy weights. Last, the fuzzy weighted method is used for the calculating results of each phosphorous partition ratio model to estimate the slag-metal phosphorous partition ratio. The simulation result of the model using the practical data from a steel plant proves the effectiveness of this dephosphorization model.
In the aspect of oxygen blowing model, this paper uses case-based reasoning approach to calculate oxygen volume blowing, which mainly contains the processes of case retrieval and case adjust. In the case retrieval process, focusing on the limitation that traditional bayesian rough set model theory can only deal with the situation of two decision classes, a improved bayesian rough set model is proposed to deal with the problem of multiple decision classes. On this condition, a y dependency function is defined to evaluate the condition attributes significance to decision attributes, and it is proved that the y function is monotonic increase with condition attributes. In the end, an algorithm to compute attribute weight is proposed based on the monotonic property of y dependency function, using which to determine the weights of factors affecting oxygen blowing. On this basis, the most nearest neighbor method is used to retrieve a group of similar cases as the current furnace occasion. In the case adjust process, it trains the hierarchical mixture of experts model with this group of similar cases, then particle swarm optimization algorithm is applyed to optimize the model parameters. Finally, calculate oxygen volume blowing in current production condition. The simulation result of the model using the practical data from a steel plant proves the effectiveness of this oxygen blowing model.
引文
[1]王新华.钢铁冶金[M].北京:高等教育出版社,2005.
[2]陶钧,谢书明,柴天佑.转炉炼钢控制模型的研究与展望[J].钢铁,1999,34(8):69-72.
[3]史美伦,徐迎铁,刘阳春,等.现代炼钢流程冶炼工序的炉料结构研究[J].炼钢,2007,23(4):39-42.
[4]H.Suito,R.Inoue.Thermodynamic Assessment of Hot Metal and Steel Dephosphorization with MnO-containing BOF Slags[J].ISIJ International,1995,35(3):258-265.
[5]H.Suito,R.Inoue.Thermodynamic Assessment of Manganese Distribution in Hot Metal and Steel[J].ISIJ International,1995,35(3):266-271.
[6]E.T.Turkdogan.Assessment of P205 Activity Coefficients in Molten Slags[J].ISIJ International,2000,40(lO):964-970.
[7]E.T.Turkdogan.Theoretical concept on slag- oxygen sensors to measure oxide activities related to FeO,Si02,and CaO contents of steelmaking slags[J].Ironmaking and Steelmaking,2000,27(1):32-36.
[8]S.M.Jung,R.J.Fruehan.Foaming Characteristics of BOF Slags[J].ISIJ International,2000,40(4):348-355.
[9]C.M.Lee,R.J.Fruehan.Phosphorus equilibrium between hot metal and slag[J].Ironmaking and Steelmaking,2005,32(6):503-508.
[10]杨文远,郑丛杰,杨立红.大型转炉炼钢脱磷的研究[J].炼钢,2002,18(1):30-34.
[11]吴伟,马嵩,邹宗树,等.1600℃高碱性渣与钢液间磷的分配比[J].材料与冶金学报,2003,2(2):83-87.
[12]吴伟,邹宗树,马嵩,等.转炉冶炼终点炉渣氧化性和碱度的研究[J].炼钢,2004,20(3):28-32.
[13]吴伟,刘浏,邹宗树,等.冶炼中磷铁水最佳复吹模式的探讨[J].钢铁,2005,40(6):33-36.
[14]吴伟,邹宗树,郭振和,等.复吹转炉最佳成渣路线的探讨[J].钢铁研究学报,2004,16(1):21-24.
[15]吴伟,吴志宏,李永祥.150t顶底复吹转炉脱磷工艺参数的研究[J].炼钢,2005,21(2):30-33.
[16]谢书明,陶钧,柴天佑.转炉炼钢终点磷的智能预报[J].控制理论与应用,2003,20(4):555-559.
[17]Jun Tao,Shusheng Ouyang,Xin Wang.Intelligent Method For BOF Endpoint[P]&[Mn]Estimation.Proceedings of the 6th World Congress on Intelligent Control and Automation[C],2006,6:21-23,Dalian,China.
[18]P.Misra,B.Deo,R.P.Chhabra.Dynamic Model of Slag Foaming in Oxygen Steelmaking Converters[J].ISIJ International,1998,38(11):1225-1232.
[19]刘跃,刘浏,佟溥翘,等.优质高碳钢高拉碳前期脱磷过程控制[J].炼钢,2006,22(2):27-29.
[20]Susumu Mukawa,Yoshimasa Mizukami.Eftect of Stirring Energy and Rate of OxygenSupply on the Rate of Hot Metal Dephosphorization[J].ISIJ International,1995,35(11):1374-1380.
[21]康复,陆志新,蒋晓放,等.宝钢BRP技术的研究与开发[J].钢铁,2005,40(3):25-28.
[22]李峻,曾加庆,高建军,等.低碱度渣铁水预处理脱磷研究[J].钢铁,2006,41(8):29-33.
[23]L.Zhang,J.Zhi,F.Mei.Basic oxygen furnace based steelmaking processes and cleanliness control at Baosteel[J].Ironmaking and Steelmaking,2006,33(2):129-139.
[24]Surlnder S Ghag,Peter C.Haye,Hae Geon Lee.Model Developmentof Slag Foaming[J].ISIJ International,1998,38(11):1208-1215.
[25]Yoshiei Kato,Haruji Okuda.Reaction Model for Carbon,Manganese,and Oxygen in Bottom Blowing with Mixed Gas in Final Stage of Steel Refining in Converter[J].ISIJ International,2003,43(11):1710-1714.
[26]C.Kubat,H.Taskin,R.Artir,A.Yilmaz.Bofy-fuzzy logic control for the basic oxygen furnace(BOF)[J].Robotics and Autonomous Systems,2004,49.193-205.
[27]刘剑,袁守谦,李建国.用基于统计模式识别的优化BP网络预报转炉的终点磷含量[J].钢铁研究学报,2005,17(3):65-68.
[28]A.M.Frattini Fileti,T.A.Pacianotto.Pitasse Cunha.Neural modeling helps the BOS process to achieve aimed end-point conditions in liquid steel[J].Engineering Applications of Artificial Intelligence,2006,19:9-17.
[29]陈忠伟,袁守谦.LD转炉冶炼的静态数学模型及实现[J].炼钢,2000,16(5):31-34.
[30]常立忠,李正邦.基于BP神经网络的转炉静态模型[J].炼钢,2006,22(6):41-43.
[31]黄金侠,金宁德.转炉冶炼终点静态控制预测模型[J].炼钢,2006,22(1):45-48.
[32]关春立,王明春,刚占库.通钢120t转炉实现计算机自动炼钢的实践[J].钢铁,2007,42(5):82-86.
[33]高新波.模糊聚类分析及其应用[M].西安:西安电子科技大学出版社,2004.
[34]Dubois.D,Prade.H.Rough fuzzy sets and fuzzy rough sets[J].International Journal of General Systems,1990,17(2-3):191-209.
[35]Pal.S.K,Mitra.P.Case Generation Using Rough Sets with Fuzzy Representation[J].IEEE Transactions on Knowledge and Data Engineering,2004,16(3):293-300.
[36]Jensen.R,Shen.Q.Fuzzy-rough attribute reduction with application to web categorization[J].Fuzzy Sets and Systems,2004,141(3):469-485.
[37]Greco.S,Inuiguchi.M,Slowinski.R.Fuzzy rough sets and multiple-premise gradual decision rules[J].International Journal of Approximate Reasoning,2006,41(2):179-211.
[38]Ziarko.W.Variable precision rough set model[J].Journal of Computer and System Sciences,1993,46(1):39-59.
[39]Slezak.D,Ziarko.W.The investigation of the Bayesian rough set model[J].International Journal of Approximate Reasoning,2005,40(1):81-91.
[40]Ziarko.W.Probabilistic approach to rough sets[J].International Journal of Approximate Reasoning,2008,49(2):272-284.
[41]Greco.S,Matarazzo.B,Slowinski.R.Parameterized rough set model using rough membership and Bayesian confirmation measures[J].International Journal of Approximate Reasoning,2008,49(2):285-300.
[42]Pawlak.Z,Skowron.A.Rudiments of rough sets[J].Information Sciences,2007,177(1):3-27.
[43]Pawlak.Z,Skowron.A.Rough sets and Boolean reasoning[J].Information Sciences,2007,177(1):41-73.
[44]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001.
[45]Xie.X.L,Beni.G.A validity measure for fuzzy clustering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1991,13(8):841-847.
[46]Tom M.Mitchell.机器学习[M].北京:机械工业出版社,2003.
[47]Simon Haykin.神经网络原理[M].北京:机械工业出版社,2006.
[48]R.E.Schapire,M.Rochery,M.Rahim,N.Gupta.Boosting With Prior Knowledge for Call Classification[J].IEEE transaction on speech and audio processing,2005,13:174-181.
[49]Chun-Xia Zhang,J.S.Zhang.RotBoost:A technique for combining Notation Forest and AdaBoost[J].Pattern Recognition Letters,2008,29:1524-1536.
[50]付忠良.关于AdaBoost有效性的分析[J].计算机研究与发展,2008,45(10):1747-1755.
[51]Y.Freund,R.E.Schapire.A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting[J].Journal of computer and system sciences,1997,55:119-139.
[52]唐伟,周志华.基于Bagging的选择性聚类集成[J].软件学报,2005,16(4):496-502.
[53]徐新卫,曹永忠,谢强,等.基于分层混合专家神经网络的Web服务失效检测机制[J].信息与控制,2007,36(6):684-689.
[54]M.I.Jordan,R.A.Jacobs.Hierarchical mixtures of experts and the EMalgorithm[J].Neural Computation,1994,6(2),181-214.
[55]谢晓锋,张文俊,杨之廉.微粒群算法综述[J].控制与决策.2003,18(2):129-134.
[2]陶钧,谢书明,柴天佑.转炉炼钢控制模型的研究与展望[J].钢铁,1999,34(8):69-72.
[3]史美伦,徐迎铁,刘阳春,等.现代炼钢流程冶炼工序的炉料结构研究[J].炼钢,2007,23(4):39-42.
[4]H.Suito,R.Inoue.Thermodynamic Assessment of Hot Metal and Steel Dephosphorization with MnO-containing BOF Slags[J].ISIJ International,1995,35(3):258-265.
[5]H.Suito,R.Inoue.Thermodynamic Assessment of Manganese Distribution in Hot Metal and Steel[J].ISIJ International,1995,35(3):266-271.
[6]E.T.Turkdogan.Assessment of P205 Activity Coefficients in Molten Slags[J].ISIJ International,2000,40(lO):964-970.
[7]E.T.Turkdogan.Theoretical concept on slag- oxygen sensors to measure oxide activities related to FeO,Si02,and CaO contents of steelmaking slags[J].Ironmaking and Steelmaking,2000,27(1):32-36.
[8]S.M.Jung,R.J.Fruehan.Foaming Characteristics of BOF Slags[J].ISIJ International,2000,40(4):348-355.
[9]C.M.Lee,R.J.Fruehan.Phosphorus equilibrium between hot metal and slag[J].Ironmaking and Steelmaking,2005,32(6):503-508.
[10]杨文远,郑丛杰,杨立红.大型转炉炼钢脱磷的研究[J].炼钢,2002,18(1):30-34.
[11]吴伟,马嵩,邹宗树,等.1600℃高碱性渣与钢液间磷的分配比[J].材料与冶金学报,2003,2(2):83-87.
[12]吴伟,邹宗树,马嵩,等.转炉冶炼终点炉渣氧化性和碱度的研究[J].炼钢,2004,20(3):28-32.
[13]吴伟,刘浏,邹宗树,等.冶炼中磷铁水最佳复吹模式的探讨[J].钢铁,2005,40(6):33-36.
[14]吴伟,邹宗树,郭振和,等.复吹转炉最佳成渣路线的探讨[J].钢铁研究学报,2004,16(1):21-24.
[15]吴伟,吴志宏,李永祥.150t顶底复吹转炉脱磷工艺参数的研究[J].炼钢,2005,21(2):30-33.
[16]谢书明,陶钧,柴天佑.转炉炼钢终点磷的智能预报[J].控制理论与应用,2003,20(4):555-559.
[17]Jun Tao,Shusheng Ouyang,Xin Wang.Intelligent Method For BOF Endpoint[P]&[Mn]Estimation.Proceedings of the 6th World Congress on Intelligent Control and Automation[C],2006,6:21-23,Dalian,China.
[18]P.Misra,B.Deo,R.P.Chhabra.Dynamic Model of Slag Foaming in Oxygen Steelmaking Converters[J].ISIJ International,1998,38(11):1225-1232.
[19]刘跃,刘浏,佟溥翘,等.优质高碳钢高拉碳前期脱磷过程控制[J].炼钢,2006,22(2):27-29.
[20]Susumu Mukawa,Yoshimasa Mizukami.Eftect of Stirring Energy and Rate of OxygenSupply on the Rate of Hot Metal Dephosphorization[J].ISIJ International,1995,35(11):1374-1380.
[21]康复,陆志新,蒋晓放,等.宝钢BRP技术的研究与开发[J].钢铁,2005,40(3):25-28.
[22]李峻,曾加庆,高建军,等.低碱度渣铁水预处理脱磷研究[J].钢铁,2006,41(8):29-33.
[23]L.Zhang,J.Zhi,F.Mei.Basic oxygen furnace based steelmaking processes and cleanliness control at Baosteel[J].Ironmaking and Steelmaking,2006,33(2):129-139.
[24]Surlnder S Ghag,Peter C.Haye,Hae Geon Lee.Model Developmentof Slag Foaming[J].ISIJ International,1998,38(11):1208-1215.
[25]Yoshiei Kato,Haruji Okuda.Reaction Model for Carbon,Manganese,and Oxygen in Bottom Blowing with Mixed Gas in Final Stage of Steel Refining in Converter[J].ISIJ International,2003,43(11):1710-1714.
[26]C.Kubat,H.Taskin,R.Artir,A.Yilmaz.Bofy-fuzzy logic control for the basic oxygen furnace(BOF)[J].Robotics and Autonomous Systems,2004,49.193-205.
[27]刘剑,袁守谦,李建国.用基于统计模式识别的优化BP网络预报转炉的终点磷含量[J].钢铁研究学报,2005,17(3):65-68.
[28]A.M.Frattini Fileti,T.A.Pacianotto.Pitasse Cunha.Neural modeling helps the BOS process to achieve aimed end-point conditions in liquid steel[J].Engineering Applications of Artificial Intelligence,2006,19:9-17.
[29]陈忠伟,袁守谦.LD转炉冶炼的静态数学模型及实现[J].炼钢,2000,16(5):31-34.
[30]常立忠,李正邦.基于BP神经网络的转炉静态模型[J].炼钢,2006,22(6):41-43.
[31]黄金侠,金宁德.转炉冶炼终点静态控制预测模型[J].炼钢,2006,22(1):45-48.
[32]关春立,王明春,刚占库.通钢120t转炉实现计算机自动炼钢的实践[J].钢铁,2007,42(5):82-86.
[33]高新波.模糊聚类分析及其应用[M].西安:西安电子科技大学出版社,2004.
[34]Dubois.D,Prade.H.Rough fuzzy sets and fuzzy rough sets[J].International Journal of General Systems,1990,17(2-3):191-209.
[35]Pal.S.K,Mitra.P.Case Generation Using Rough Sets with Fuzzy Representation[J].IEEE Transactions on Knowledge and Data Engineering,2004,16(3):293-300.
[36]Jensen.R,Shen.Q.Fuzzy-rough attribute reduction with application to web categorization[J].Fuzzy Sets and Systems,2004,141(3):469-485.
[37]Greco.S,Inuiguchi.M,Slowinski.R.Fuzzy rough sets and multiple-premise gradual decision rules[J].International Journal of Approximate Reasoning,2006,41(2):179-211.
[38]Ziarko.W.Variable precision rough set model[J].Journal of Computer and System Sciences,1993,46(1):39-59.
[39]Slezak.D,Ziarko.W.The investigation of the Bayesian rough set model[J].International Journal of Approximate Reasoning,2005,40(1):81-91.
[40]Ziarko.W.Probabilistic approach to rough sets[J].International Journal of Approximate Reasoning,2008,49(2):272-284.
[41]Greco.S,Matarazzo.B,Slowinski.R.Parameterized rough set model using rough membership and Bayesian confirmation measures[J].International Journal of Approximate Reasoning,2008,49(2):285-300.
[42]Pawlak.Z,Skowron.A.Rudiments of rough sets[J].Information Sciences,2007,177(1):3-27.
[43]Pawlak.Z,Skowron.A.Rough sets and Boolean reasoning[J].Information Sciences,2007,177(1):41-73.
[44]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001.
[45]Xie.X.L,Beni.G.A validity measure for fuzzy clustering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1991,13(8):841-847.
[46]Tom M.Mitchell.机器学习[M].北京:机械工业出版社,2003.
[47]Simon Haykin.神经网络原理[M].北京:机械工业出版社,2006.
[48]R.E.Schapire,M.Rochery,M.Rahim,N.Gupta.Boosting With Prior Knowledge for Call Classification[J].IEEE transaction on speech and audio processing,2005,13:174-181.
[49]Chun-Xia Zhang,J.S.Zhang.RotBoost:A technique for combining Notation Forest and AdaBoost[J].Pattern Recognition Letters,2008,29:1524-1536.
[50]付忠良.关于AdaBoost有效性的分析[J].计算机研究与发展,2008,45(10):1747-1755.
[51]Y.Freund,R.E.Schapire.A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting[J].Journal of computer and system sciences,1997,55:119-139.
[52]唐伟,周志华.基于Bagging的选择性聚类集成[J].软件学报,2005,16(4):496-502.
[53]徐新卫,曹永忠,谢强,等.基于分层混合专家神经网络的Web服务失效检测机制[J].信息与控制,2007,36(6):684-689.
[54]M.I.Jordan,R.A.Jacobs.Hierarchical mixtures of experts and the EMalgorithm[J].Neural Computation,1994,6(2),181-214.
[55]谢晓锋,张文俊,杨之廉.微粒群算法综述[J].控制与决策.2003,18(2):129-134.