噪声特性的回归模型及其在短期风速预报中的应用
|
摘要
回归模型注重挖掘数据的线性或非线性结构规律,它为研究人员和工程师利用数据进行学习提供了强大的工具.回归模型已成功应用于研究和应用等各个领域,包括社会科学、经济学、金融学、电网运行的风功率预测等.
经典的核岭回归模型和基于高斯噪声特性的-支持向量回归模型都假设噪声特性服从高斯分布.然而在许多实际应用领域中,如风功率预报、相关电磁波的到达方向估计问题等,噪声特性不服从高斯分布,而服从Beta分布、拉普拉斯分布,或者别的分布.此时,经典的回归技术不是最优的.利用Bayesian方法,得到一般噪声特性的损失函数,构造了新的基于噪声特性的核岭回归模型(GN-KRR)框架结构;提出了新的基于噪声特性的-支持向量回归模型(GN-SVR)通用框架.
基于高斯噪声特性的核岭回归模型(GN-KRR)、-支持向量回归模型(GN-SVR)都假设噪声特性服从均值为0、同方差2的高斯分布.作者利用持续法统计得到,风速预报误差不服从均值为0、同方差2的高斯分布,而服从均值为0,异方差2i(i1,2,, l)的高斯分布.此时模型GN-KRR和GN-SVR不是最优的.利用异方差噪声特性的损失函数,提出了新的基于异方差噪声特性-支持向量回归模型(HGN-SVR)框架结构.
在ε-支持向量回归和粗糙-支持向量回归模型的基础上,研究了新的粗糙-支持向量回归模型.利用固定对称边界粗糙-不敏感损失函数,构造固定对称边界粗糙ε-支持向量回归模型;利用固定非对称边界粗糙-不敏感损失函数,构造固定非对称边界粗糙-支持向量回归模型.通过引进拉格朗日函数和根据KKT条件,得到了粗糙-支持向量回归模型的对偶问题.
作者根据Karush-Kuhn-Tucker (KKT)条件,通过构造拉格朗日泛函,得到了模型GN-KRR、 GN-SVR、 HGN-SVR的对偶问题.利用增广拉格朗日乘子法求解模型GN-KRR、GN-SVR的最优解,利用随机梯度下降法求解模型HGN-SVR的最优解.将上述三类模型应用于短期风速预报中,实验结果表明提出的噪声特性回归模型的有效性.
Regression is an old topic in the domain of learning functions from a set of samples. Itprovides researchers and engineers with a powerful tool to extract hidden rules of data. Thetrained model is used to predict future events with the information of past or present events.Regression analysis is now successfully applied in nearly all fields of science and technology,including the social sciences, economics, finance, wind power prediction for grid operation.However, this domain is still attracting much attention from research and application domains.The main research contents of this work are as follows.
The classical kernel ridge regression (KRR) and-Support vector regression (-SVR)techniques are aimed at discovering a linear or nonlinear structure hidden in original data.They make an assumption that the noise distribution is Gaussian. However, it is reported thatthe noise models in some real-world practical applications, such as wind power forecastingand direction of the direction-of-arrival of coherent electromagnetic waves impingingestimation problem, do not satisfy Gaussian distribution, but beta distribution, Laplaciandistribution, or other models. In these cases the current regression techniques are not optimal.Using Bayesian approach, we derive an optimal loss function for a general noise model. Wepropose a new framework of kernel ridge regression for the general noise mode (N KRR)and a novel-support vector regression model for the general noise model (N-SVR),respectively.
The classical GN KRR and GN-SVRtechniques take an assumption that the noiseis Gaussian with zero mean and the same variance. However, it is found that the noise modelsin some practical applications satisfy Gaussian distribution with zero mean andheteroscedasticity, such as wind speed forecasting. In this case, the derived models are notoptimal. Using the optimal loss function for heteroscedastic noise model, propose a newframework of-SVRfor heteroscedastic noise model (HN-SVR).
Two new rough-support vector regression models are proposed based on-supportvector regression, rough-support vector regression and rough set theory. Firstly, a roughboundary-insensitive tube is defined with fixed symmetrical boundary rough
ε-insensitive loss function and the method of optimization and-support vector regressionmodel. Then we design a fixed symmetrical boundary rough-support vector regressionmodel (RFSM ε-SVR). Secondly, we extend the model to the case that asymmetrical lossfunction is considered. Finally, according to Karush-Kuhn-Tucker (KKT) conditions, wederive their dual problems by introducing the Lagrangian functional into rough-supportvector regression models.
According to Karush-Kuhn-Tucker (KKT) conditions, we derive their dual problems byintroducing the Lagrangian functional into N-KRR, N-SVRand HN-SVR. The AugmentedLagrangian Multiplier (ALM) method is applied to solve the dual models of the modelsN-KRR and N-SVR. The Stochastic Gradient Descent (SGD) method is applied to solvethe model HGN-SVR. We test the proposed techniques to short-term wind speed prediction.Experimental results confirm the effectiveness of the proposed models.
经典的核岭回归模型和基于高斯噪声特性的-支持向量回归模型都假设噪声特性服从高斯分布.然而在许多实际应用领域中,如风功率预报、相关电磁波的到达方向估计问题等,噪声特性不服从高斯分布,而服从Beta分布、拉普拉斯分布,或者别的分布.此时,经典的回归技术不是最优的.利用Bayesian方法,得到一般噪声特性的损失函数,构造了新的基于噪声特性的核岭回归模型(GN-KRR)框架结构;提出了新的基于噪声特性的-支持向量回归模型(GN-SVR)通用框架.
基于高斯噪声特性的核岭回归模型(GN-KRR)、-支持向量回归模型(GN-SVR)都假设噪声特性服从均值为0、同方差2的高斯分布.作者利用持续法统计得到,风速预报误差不服从均值为0、同方差2的高斯分布,而服从均值为0,异方差2i(i1,2,, l)的高斯分布.此时模型GN-KRR和GN-SVR不是最优的.利用异方差噪声特性的损失函数,提出了新的基于异方差噪声特性-支持向量回归模型(HGN-SVR)框架结构.
在ε-支持向量回归和粗糙-支持向量回归模型的基础上,研究了新的粗糙-支持向量回归模型.利用固定对称边界粗糙-不敏感损失函数,构造固定对称边界粗糙ε-支持向量回归模型;利用固定非对称边界粗糙-不敏感损失函数,构造固定非对称边界粗糙-支持向量回归模型.通过引进拉格朗日函数和根据KKT条件,得到了粗糙-支持向量回归模型的对偶问题.
作者根据Karush-Kuhn-Tucker (KKT)条件,通过构造拉格朗日泛函,得到了模型GN-KRR、 GN-SVR、 HGN-SVR的对偶问题.利用增广拉格朗日乘子法求解模型GN-KRR、GN-SVR的最优解,利用随机梯度下降法求解模型HGN-SVR的最优解.将上述三类模型应用于短期风速预报中,实验结果表明提出的噪声特性回归模型的有效性.
Regression is an old topic in the domain of learning functions from a set of samples. Itprovides researchers and engineers with a powerful tool to extract hidden rules of data. Thetrained model is used to predict future events with the information of past or present events.Regression analysis is now successfully applied in nearly all fields of science and technology,including the social sciences, economics, finance, wind power prediction for grid operation.However, this domain is still attracting much attention from research and application domains.The main research contents of this work are as follows.
The classical kernel ridge regression (KRR) and-Support vector regression (-SVR)techniques are aimed at discovering a linear or nonlinear structure hidden in original data.They make an assumption that the noise distribution is Gaussian. However, it is reported thatthe noise models in some real-world practical applications, such as wind power forecastingand direction of the direction-of-arrival of coherent electromagnetic waves impingingestimation problem, do not satisfy Gaussian distribution, but beta distribution, Laplaciandistribution, or other models. In these cases the current regression techniques are not optimal.Using Bayesian approach, we derive an optimal loss function for a general noise model. Wepropose a new framework of kernel ridge regression for the general noise mode (N KRR)and a novel-support vector regression model for the general noise model (N-SVR),respectively.
The classical GN KRR and GN-SVRtechniques take an assumption that the noiseis Gaussian with zero mean and the same variance. However, it is found that the noise modelsin some practical applications satisfy Gaussian distribution with zero mean andheteroscedasticity, such as wind speed forecasting. In this case, the derived models are notoptimal. Using the optimal loss function for heteroscedastic noise model, propose a newframework of-SVRfor heteroscedastic noise model (HN-SVR).
Two new rough-support vector regression models are proposed based on-supportvector regression, rough-support vector regression and rough set theory. Firstly, a roughboundary-insensitive tube is defined with fixed symmetrical boundary rough
ε-insensitive loss function and the method of optimization and-support vector regressionmodel. Then we design a fixed symmetrical boundary rough-support vector regressionmodel (RFSM ε-SVR). Secondly, we extend the model to the case that asymmetrical lossfunction is considered. Finally, according to Karush-Kuhn-Tucker (KKT) conditions, wederive their dual problems by introducing the Lagrangian functional into rough-supportvector regression models.
According to Karush-Kuhn-Tucker (KKT) conditions, we derive their dual problems byintroducing the Lagrangian functional into N-KRR, N-SVRand HN-SVR. The AugmentedLagrangian Multiplier (ALM) method is applied to solve the dual models of the modelsN-KRR and N-SVR. The Stochastic Gradient Descent (SGD) method is applied to solvethe model HGN-SVR. We test the proposed techniques to short-term wind speed prediction.Experimental results confirm the effectiveness of the proposed models.
引文
[1] Lange M.On the uncertainty of wind power predictions-analysis of the forecast accuracyand statistical distribution of errors[M]. J. Sol. Energy Eng.,2005,127:177-184.
[2]中国新能源与可再生能源网http://www.crein.org.cn/.
[3]中国电力科学研究院http://www.epri.sgcc.com.cn/.
[4]宗培书,汤剑平,潘益农,许春艳.江苏东台地区风速风能特征分析及高精度模拟[J].南京大学学报(自然科学版),2013,49(3):300-310.
[5]杨志凌.风电场功率短期预测方法优化的研究[D].北京:华北电力大学,2011,6.
[6] Hoven I V. Power spectrum of horizontal wind speed in the frequency range from0.0007to900cycles per hour[J]. Journal of Meteorology,1957,14(2):160-164.
[7] Fabbri A, Román T G S, Abbad J R, Quezada V H M. Assessment of the cost associatedwith wind generation prediction errors in a liberalized electricity market[J]. IEEE Trans.on power systems,2005,20(3):1440-1446.
[8] Bludszuweit H, Antonio J, Llombart A. Statistical analysis of wind power forecasterror[J]. IEEE Trans. on power systems,2008,23(3):983-991.
[9] Hoerl A E. Application of ridge analysis to regression problems[J]. Chemicalengineering progress,1962,58:54-59.
[10] Hoerl A E, Kennard R W. Ridge regression: biased estimation for nonorthogoallproblems[J]. Technometrics,1970,12(1):55-67.
[11] Hastie T, Tibshirani R, Friedman J H. The elements of statistical learning[M]. New York:Springer,2001.
[12] Hastie T, Tibshirani R, Friedman J. The elements of statistical learning: data mining,inference, and prediction[M]. New York: Springer,2009.
[13] Tikhonov A A, Arsenin V Y. Solutions of ill-posed problems[M]. New York: Wiley,1977.
[14] Saunders C, Gammerman A, Vovk V. Ridge regression learning algorithm indualvariables[C].In Proceedings of the15th international conference on machinelearning, Morgan Kaufmann,1998:515-521.
[15] Suykens J A K, Vandewalle J. Least squares support vector machine classifiers[J].Neural processing letters,1999,9:293-300.
[16] Suykens J A K, Lukas L, Vandewalle J. Sparse approximation using least square vectormachines[J]. IEEE Proc. Int. Symp. Circuits Syst., Genvea, Switzerland,2000:757-760.
[17] Suykens J A K, Van Gestel T, De Brabanter J, et al. Least squares support vectormachines[M]. Singapore: World Scientific,2002.
[18] Vapnik V. The nature of statistical learning theory[M]. New York: Springer Verlag,1995.
[19] Cortes C, Vapnik V. Support vector networks[J]. Machines learning,1995,20(3):273-297.
[20]张学工译.统计学习理论的本质(中文版).北京:清华大学出版社,2000.
[21] Vapnik V, Golowich S E, Smola A. Support vector method for function approximation,regression estimation, and signal processing[C]. In advances in neural informationprocessings systems9. Cambridge, MA: The MIT Press,1996:281-287.
[22] Vapnik V, Statistical learning theory[M]. New York: Wiley,1998.
[23] Burges C J C. A tutorial on support machines for pattern recognition[J]. Data miningknowledge discovery,1998,2(2):121-167.
[24] Smola A, Sch错误！未找到引用源。lkopf B. A tutorial on support regression[J].Statistics and computation,2004,14(3):199-222.
[25] Olvi L, David R M. Robust linear and support vector regression[J]. IEEE Trans. onpattern analysis and machine intelligence,2000,22(9):950-955.
[26] James T, Kwok I, Tsang W. Linear dependency between ε and the input noise inε-support vector regression[J]. IEEE Trans. on neural networks,2003,14(3):544-553.
[27] Wei C, Sathiya K, Chong Jin O. Bayesian support vector regression using a unified lossfunction[J]. IEEE Trans. on neural networks,2004,15(1):544-553.
[28] Sch错误！未找到引用源。lkopf B, Smola A J, Williamson R C, et al. New supportvector algorithms [J]. Neural computation,2000,12:1207-1245.
[29] Chalimourda A, Sch错误！未找到引用源。lkopf B, Smola A J. Experimentally optimalin support vector regression for different noise models and parameter settings[J].Neural networks,2004,17(1):127-141.
[30] Chih C C, Chih J L. Training v-support vector regression: theory and algorithms[J].Neural computation,2002,14:1959-1977.
[31] Sch错误！未找到引用源。lkopf B, Smola A J. Learning with kernels: support vectormachines, regularization,optimization, and beyond [M]. Cambridge, MA: MIT Press,2002.
[32] Wu Q. A hybrid-forecasting model based on Gaussian support vector machine andchaotic particle swarm optimization[J]. Expert systems with applications,2010,37:2388-2394.
[33] Wu Q, Rob Law. The forecasting model based on modified SVRM and PSO penalizingGaussian noise[J]. Expert systems with applications,2011,38(3):1887-1894.
[34] Bofinger S, Luig A, Beyer H G. Qualification of wind power forecasts[C]. In Proc.global wind power Conf. GWPC'02, Paris, France,2002.
[35] Canavos G C. Applied probability and statistical methods[M]. Little Brown andCompany,1984.
[36] Bishop C M. Pattern recognition and machine learning[M]. New York: Springer,2006:71-73.
[37]陈希儒,倪国煕.数理统计学教程[M].合肥:中国科技大学出版社,2009.
[38] Zhang Y, Wan Q, Zhao H P and Yang W L. Support vector regression for basis selectionin Laplacian noise environment[J]. IEEE signal letters,2007,14(11):871-874.
[39] Randazzo A, Abou-Khousa M A, Pastorino M, et al. Direction of arrival estimationbased on support vector regression: experimental validation and comparison withmusic[J].IEEE antennas and wireless propagation letters,2007,6:379-382.
[40]谢宇.回归分析[M].北京:社会科学文献出版社,2010.
[41] Maekensen R, Lange B, Schlogl F. Integrating wind energy into public power suppysystems-German state of the art[C]. In: proceedings of the second internationalconference on integration of renewable and distributed energy resourees, NaPa, CA,USA,4-8, December2006:12.
[42] Bailey B, Stewart R. Wind forecasting for wind power stations[C]. In: proceedings ofwind energy conversion, Edinburgh, UK,1987:265-269.
[43] Wegley H, Formica W. Test applications of a semi-objective approach to windforecasting for wind energy applications[C]. R PNL-4403, Pacific NorthwestLaboratory,1983.
[44] Alexiadis M C, Dikopoulos P S, Sahsamanoglou H S, et al. Shortterm foreasting of windspeed and related electrical power[J]. Sol. Energy1998,63(1):61-68.
[45] Focken U, Lange M, Monnich K, et al. Short-term prediction of the aggregated poweroutput of wind farms e a statistical analysis of the reduction of the prediction error byspatial smoothing effects[J]. Wind Eng. Ind. Aerodyn,2002,90:231-246.
[46] Barbounis T G, Theocharis J B. A locally recurrent fuzzy neural network withapplication to the wind speed prediction using spatial correlation[J]. Neurocomputing,2007,70:1525-1542.
[47] Barbounis T G, Theocharis J B. Locally recurrent neural networks for wind speedprediction using spatial correlation[J]. Information science2007,177:5775-5797.
[48] Negnevitsky M, Potter C W. Innovative short-term wind generation predictiontechniques[C]. In: proceedings of the power systems conference and exposition,2006:60-65.
[49] Landberg L. A mathematical look at a physical power prediction model[J]. Wind energy,1998, l(l):23-28.
[50] Notis C, Trettel D, Aquino J, Piazza T,et al. Learning to forecast wind at remote sites forwind energy applications[C]. R PNL-4318, Pacific Northwest Laboratry,1983.
[51] Wegley H, Kosorok M, Formica W. Sub-hourly wind forecasting techniques for windturbine operations[C]. R PNL-4894, Pacific Northwest Laboratory,1984.
[52] Geerts H. Short range Prediction of wind speeds: a system-theoretic approach[C]. In:proceedings of European wind energy conference, Hamburg, Germany,1984:594-599.
[53] Torres J L, Garcia A, De Blas M, et al. Forecast of hourly average wind speed withARMA models in Navarre (Spain)[J]. Sol. Energy,2005,79(1):65-77.
[54] Kavasseri R G, Seetharaman K. Day-ahead wind speed forecasting using f-ARIMAmodels[M]. Renew energy,2009,34(5):1388-1393.
[55] Li G, Shi J. On comparing three artificial neural networks for wind speed forecasting[J].Applied energ2010,87(7):2313-2320.
[56] Cadenas E, Rivera W. Short term wind speed forecasting in La Venta, Mexico usingartificial neural networks[J]. Renew energy,2009,34(1):274-278.
[57] Barbounis T G, Theocharis J B. Locally recurrent neural networks for long-term windspeed and power prediction[J]. Neurocomputing2006,69(46):466-496.
[58] Catalao J P S, Pousinho H M I, Mendes V M F. An artificial neural network approachfor short-term wind power forecasting in Portugal[C]. IEEE Conf.15th internationalconference on intelligent system applications to power systems,2009:1-5.
[59] Ortiz-Garcia E G, Salcedo-Sanz S, Perez-Bellido A M, et al. Portilla-Figueras JA, PrietoL. Short-term wind speed prediction in wind farms based on banks of support vectormachines[J].Wind energy,2011,14(2):193-207.
[60] Mohandes M A, Halawani T O,Rehman S,et al. Support vector machines for wind speedprediction[J]. Renew energy,2004,29(6):939-947.
[61] Sancho S S, Emilio G O, et al. Short term wind speed prediction based on evolutionarysupport vector regression algorithms[J]. Expert systems with applications2011,38:4052–4057.
[62] Zhou J Y, Shi J, Li G. Fine tuning support vector machines for short-term wind speedforecasting[J]. Energy conversion and management2011,52:1990-1998.
[63] Liu H, Tian H Q, Chen C, Li Y F. A hybrid statistical method to predict wind speed andwind power[J]. Renew energy,2010,35(8):1857-1861.
[64] Catalao J P S, Pousinho H M I, Mendes V M F. Short-term wind power forecasting inPortugal by neural networks and wavelet transform[J]. Renew energ2011,36(4):1245-1251.
[65] Girosi F. Models of noise and robust estimates. A. I. Memo1287, Artificial intelligencelaboratory, Massachusetts institute of technology,1991.
[66] Pontil M, Mukherjee S and Girosi F. On the noise model of support vector machinesregression. A. I. Memo1651, Artificial intelligence laboratory, Massachusetts instituteof technology,1998.
[67] Pontil M, Mukherjee S, Girosi F. On the noise model of support vector machinesregression [J]. ALT,2000, LNAI:316-324.
[68] Chu W, Keerthi S S, Ong C J. Bayesian support vector regression using a unified lossfunction [J]. IEEE Trans.on neural networks,2004,22(1):29-44.
[69] Klaus-Robert M, Sebastian M. An introduction to kernel-based learning algorithms [J].IEEE Trans. on neural networks,2001,12(2):181-202.
[70] Huber P J. Robust estimation of a location parameter [J]. The annals of mathematicalstatistics,1964,35(1):73-101.
[71] Huber P J. Robust statistics [M]. New York: Wiley,1981.
[72] Fan S, Liao J R, et al. Forecasting the wind generation using a two-stage network basedon meteorological information [J]. IEEE Trans. on energy conversion,2009,24(2):474-482.
[73] Guo Z H, Zhao J, Zhang W Y, Wang J Z. A corrected hybrid approach for wind speedprediction in Hexi Corridor of China [J]. Energy,2011,36:1668-1679.
[74]邓乃杨,田英杰.数据挖掘中的新方法:支持向量机[M].科学出版社,2004,6.
[75]邓乃杨,田英杰.支持向量机:理论、算法与拓展[M].科学出版社,2009,8.
[76]田英杰.支持向量回归机及其应用研究[D].北京:中国农业大学,2005,6.
[77] Rockafellar R T. The multiplier method of hestenes and powell applied to convexprogramming[J]. Journal of optimization theory and applixations,1973,12(6):555-562.
[78] Rockafellar R T. Augmented Lagrange multiplier functions and duality in non-convexprogramming[J]. SIAM Journal on control,1974,12(2),268-285.
[79] Boyd S, Vandenberghe L. Convex optimization[M]. Cambridge University Press,2004.
[80]马昌凤.最优化方法及其Matlab程序设计.科学出版社,2010.8.
[81]杨志民,刘广利.不确定性支持向量机原理及应用[M].科学出版社,2007,4.
[82]杨志民,刘广利.不确定性支持向量机:算法及应用[M].科学出版社,2012,1.
[83] Bottou L. Stochastic gradient descent on toy problems,2007. http://leon.bottou.org/projects/sgd.
[84] Bottou L. Large-scale machine learning with stochastic gradient descent[C]. Proceed-ings of the19th international conference on computational statistics(COMPSTAT'2010),Paris, Springer, August2010,177–187.
[85] Bordes A, Bottou L, Gallinari P. SGD-QN: careful quasi Newton stochastic gradientdescent. Journal of machine learning research,2009,10:1737-1754.
[86] Wang Z, Crammer K, Vucetic W. Breaking the curse of kernelization: budgetedstochastic gradient descent for large-scale SVM training. Journal of machine learningresearch,2012,13:3103-3131.
[87] Pawlak Z. Rough sets [J]. International Journal of computer and information science.1982,11(5):341-356.
[88] Pawlak Z. Rough sets: theoretical aspects of reasoning of data [M]. Dordrecht: KluwerAcademic Publishers,1991.
[89] Pawlak Z, Skoworn A. Rudiments of rough sets [J]. Information science,2007,177:3-27.
[90] Bonikowaski Z., Algebraic structrues of rough sets. Fuzzy sets and knowledge discovery,Springer-Verlag, Berlin,1995:242-247.
[91] Yao Y Y. Constructive and algebraic methods of the theory of rough sets. Informationsciences,1998,109:21-47.
[92] Bonikowaski Z. Algebraic structrues of rough sets. Fuzzy sets and knowledge discovery,Springer-Verlag, Berlin,1995:242-247.
[93] Liu G L, Zhu W. The algebraic structures of generalized rough set theory. Informationsciences,2008,178:4105–4113.
[94] Zhang W X, Mi J S, Wu W Z. Approaches to knowledge reductions in inconsistentsystems. International journal of intelligent systems,2003,18:989-1000.
[95]张文修,梁怡,吴伟志.信息系统与知识发现[M].北京:科学出版社,2003,9.
[96]胡清华,于达仁.应用粗糙计算[M].北京:科学出版社,2012,6.
[97]王国胤,姚一豫,于洪.粗糙集理论与应用研究综述[J].计算机学报,2009,32(7):1229-1247.
[98] Mi J S, Wu W Z, Zhang W X. Approaches to knowledge reductions based on variableprecision rough sets model. Information sciences,2004,159:255-272.
[99] Zhao Y P, Sun J G. Rough-support vector regression [J]. Expert systems withapplications [J].2009,36(6):9793–9798.
[100] Lingras P, Butz C J. Rough support vector regression [J]. European journal ofoperational research.2010,206(2):445–455.
[101]高爽,冬雷,高阳,廖晓钟.基于粗糙集理论的中长期风速预测[J].中国电机工程学报,2012,32(1):32-37.
[102]韩虎.基于粗糙集理论的中长期风速预测[D].兰州：兰州交通大学,2011,12.
[103] Yang H, Chan L, King I. Support vector machine regression for volatile stockmarketprediction [C]. In: Yin, H, Allinson, N, Freeman, R, Keane, J, Hubbard, S (Eds.),Intelligent data engineering and automated learning. Springer, LNCS,2002,2412:391-396.
[104] Yang H. Margin variations in support vector regression for the stock market prediction.M. Phil. dissertation, department of computer science and engineering, The ChineseUniversity of Hong Kong. http://www.svms.o rg/finance/Yang2003.
[105] Fiacco, A.V. and McCormick, G.P. Nonlinear programming. Sequential unconstrainedminimization techniques. Society or industrial and applied mathematics. First publishedin1968by research analysis corporation.
[106] Chen, D.S. and Jian, R.C. A robust backpropagation learning algorithm for function ap-proximation. IEEE Trans. on neural networks,1994,5(3):467-479.
[2]中国新能源与可再生能源网http://www.crein.org.cn/.
[3]中国电力科学研究院http://www.epri.sgcc.com.cn/.
[4]宗培书,汤剑平,潘益农,许春艳.江苏东台地区风速风能特征分析及高精度模拟[J].南京大学学报(自然科学版),2013,49(3):300-310.
[5]杨志凌.风电场功率短期预测方法优化的研究[D].北京:华北电力大学,2011,6.
[6] Hoven I V. Power spectrum of horizontal wind speed in the frequency range from0.0007to900cycles per hour[J]. Journal of Meteorology,1957,14(2):160-164.
[7] Fabbri A, Román T G S, Abbad J R, Quezada V H M. Assessment of the cost associatedwith wind generation prediction errors in a liberalized electricity market[J]. IEEE Trans.on power systems,2005,20(3):1440-1446.
[8] Bludszuweit H, Antonio J, Llombart A. Statistical analysis of wind power forecasterror[J]. IEEE Trans. on power systems,2008,23(3):983-991.
[9] Hoerl A E. Application of ridge analysis to regression problems[J]. Chemicalengineering progress,1962,58:54-59.
[10] Hoerl A E, Kennard R W. Ridge regression: biased estimation for nonorthogoallproblems[J]. Technometrics,1970,12(1):55-67.
[11] Hastie T, Tibshirani R, Friedman J H. The elements of statistical learning[M]. New York:Springer,2001.
[12] Hastie T, Tibshirani R, Friedman J. The elements of statistical learning: data mining,inference, and prediction[M]. New York: Springer,2009.
[13] Tikhonov A A, Arsenin V Y. Solutions of ill-posed problems[M]. New York: Wiley,1977.
[14] Saunders C, Gammerman A, Vovk V. Ridge regression learning algorithm indualvariables[C].In Proceedings of the15th international conference on machinelearning, Morgan Kaufmann,1998:515-521.
[15] Suykens J A K, Vandewalle J. Least squares support vector machine classifiers[J].Neural processing letters,1999,9:293-300.
[16] Suykens J A K, Lukas L, Vandewalle J. Sparse approximation using least square vectormachines[J]. IEEE Proc. Int. Symp. Circuits Syst., Genvea, Switzerland,2000:757-760.
[17] Suykens J A K, Van Gestel T, De Brabanter J, et al. Least squares support vectormachines[M]. Singapore: World Scientific,2002.
[18] Vapnik V. The nature of statistical learning theory[M]. New York: Springer Verlag,1995.
[19] Cortes C, Vapnik V. Support vector networks[J]. Machines learning,1995,20(3):273-297.
[20]张学工译.统计学习理论的本质(中文版).北京:清华大学出版社,2000.
[21] Vapnik V, Golowich S E, Smola A. Support vector method for function approximation,regression estimation, and signal processing[C]. In advances in neural informationprocessings systems9. Cambridge, MA: The MIT Press,1996:281-287.
[22] Vapnik V, Statistical learning theory[M]. New York: Wiley,1998.
[23] Burges C J C. A tutorial on support machines for pattern recognition[J]. Data miningknowledge discovery,1998,2(2):121-167.
[24] Smola A, Sch错误！未找到引用源。lkopf B. A tutorial on support regression[J].Statistics and computation,2004,14(3):199-222.
[25] Olvi L, David R M. Robust linear and support vector regression[J]. IEEE Trans. onpattern analysis and machine intelligence,2000,22(9):950-955.
[26] James T, Kwok I, Tsang W. Linear dependency between ε and the input noise inε-support vector regression[J]. IEEE Trans. on neural networks,2003,14(3):544-553.
[27] Wei C, Sathiya K, Chong Jin O. Bayesian support vector regression using a unified lossfunction[J]. IEEE Trans. on neural networks,2004,15(1):544-553.
[28] Sch错误！未找到引用源。lkopf B, Smola A J, Williamson R C, et al. New supportvector algorithms [J]. Neural computation,2000,12:1207-1245.
[29] Chalimourda A, Sch错误！未找到引用源。lkopf B, Smola A J. Experimentally optimalin support vector regression for different noise models and parameter settings[J].Neural networks,2004,17(1):127-141.
[30] Chih C C, Chih J L. Training v-support vector regression: theory and algorithms[J].Neural computation,2002,14:1959-1977.
[31] Sch错误！未找到引用源。lkopf B, Smola A J. Learning with kernels: support vectormachines, regularization,optimization, and beyond [M]. Cambridge, MA: MIT Press,2002.
[32] Wu Q. A hybrid-forecasting model based on Gaussian support vector machine andchaotic particle swarm optimization[J]. Expert systems with applications,2010,37:2388-2394.
[33] Wu Q, Rob Law. The forecasting model based on modified SVRM and PSO penalizingGaussian noise[J]. Expert systems with applications,2011,38(3):1887-1894.
[34] Bofinger S, Luig A, Beyer H G. Qualification of wind power forecasts[C]. In Proc.global wind power Conf. GWPC'02, Paris, France,2002.
[35] Canavos G C. Applied probability and statistical methods[M]. Little Brown andCompany,1984.
[36] Bishop C M. Pattern recognition and machine learning[M]. New York: Springer,2006:71-73.
[37]陈希儒,倪国煕.数理统计学教程[M].合肥:中国科技大学出版社,2009.
[38] Zhang Y, Wan Q, Zhao H P and Yang W L. Support vector regression for basis selectionin Laplacian noise environment[J]. IEEE signal letters,2007,14(11):871-874.
[39] Randazzo A, Abou-Khousa M A, Pastorino M, et al. Direction of arrival estimationbased on support vector regression: experimental validation and comparison withmusic[J].IEEE antennas and wireless propagation letters,2007,6:379-382.
[40]谢宇.回归分析[M].北京:社会科学文献出版社,2010.
[41] Maekensen R, Lange B, Schlogl F. Integrating wind energy into public power suppysystems-German state of the art[C]. In: proceedings of the second internationalconference on integration of renewable and distributed energy resourees, NaPa, CA,USA,4-8, December2006:12.
[42] Bailey B, Stewart R. Wind forecasting for wind power stations[C]. In: proceedings ofwind energy conversion, Edinburgh, UK,1987:265-269.
[43] Wegley H, Formica W. Test applications of a semi-objective approach to windforecasting for wind energy applications[C]. R PNL-4403, Pacific NorthwestLaboratory,1983.
[44] Alexiadis M C, Dikopoulos P S, Sahsamanoglou H S, et al. Shortterm foreasting of windspeed and related electrical power[J]. Sol. Energy1998,63(1):61-68.
[45] Focken U, Lange M, Monnich K, et al. Short-term prediction of the aggregated poweroutput of wind farms e a statistical analysis of the reduction of the prediction error byspatial smoothing effects[J]. Wind Eng. Ind. Aerodyn,2002,90:231-246.
[46] Barbounis T G, Theocharis J B. A locally recurrent fuzzy neural network withapplication to the wind speed prediction using spatial correlation[J]. Neurocomputing,2007,70:1525-1542.
[47] Barbounis T G, Theocharis J B. Locally recurrent neural networks for wind speedprediction using spatial correlation[J]. Information science2007,177:5775-5797.
[48] Negnevitsky M, Potter C W. Innovative short-term wind generation predictiontechniques[C]. In: proceedings of the power systems conference and exposition,2006:60-65.
[49] Landberg L. A mathematical look at a physical power prediction model[J]. Wind energy,1998, l(l):23-28.
[50] Notis C, Trettel D, Aquino J, Piazza T,et al. Learning to forecast wind at remote sites forwind energy applications[C]. R PNL-4318, Pacific Northwest Laboratry,1983.
[51] Wegley H, Kosorok M, Formica W. Sub-hourly wind forecasting techniques for windturbine operations[C]. R PNL-4894, Pacific Northwest Laboratory,1984.
[52] Geerts H. Short range Prediction of wind speeds: a system-theoretic approach[C]. In:proceedings of European wind energy conference, Hamburg, Germany,1984:594-599.
[53] Torres J L, Garcia A, De Blas M, et al. Forecast of hourly average wind speed withARMA models in Navarre (Spain)[J]. Sol. Energy,2005,79(1):65-77.
[54] Kavasseri R G, Seetharaman K. Day-ahead wind speed forecasting using f-ARIMAmodels[M]. Renew energy,2009,34(5):1388-1393.
[55] Li G, Shi J. On comparing three artificial neural networks for wind speed forecasting[J].Applied energ2010,87(7):2313-2320.
[56] Cadenas E, Rivera W. Short term wind speed forecasting in La Venta, Mexico usingartificial neural networks[J]. Renew energy,2009,34(1):274-278.
[57] Barbounis T G, Theocharis J B. Locally recurrent neural networks for long-term windspeed and power prediction[J]. Neurocomputing2006,69(46):466-496.
[58] Catalao J P S, Pousinho H M I, Mendes V M F. An artificial neural network approachfor short-term wind power forecasting in Portugal[C]. IEEE Conf.15th internationalconference on intelligent system applications to power systems,2009:1-5.
[59] Ortiz-Garcia E G, Salcedo-Sanz S, Perez-Bellido A M, et al. Portilla-Figueras JA, PrietoL. Short-term wind speed prediction in wind farms based on banks of support vectormachines[J].Wind energy,2011,14(2):193-207.
[60] Mohandes M A, Halawani T O,Rehman S,et al. Support vector machines for wind speedprediction[J]. Renew energy,2004,29(6):939-947.
[61] Sancho S S, Emilio G O, et al. Short term wind speed prediction based on evolutionarysupport vector regression algorithms[J]. Expert systems with applications2011,38:4052–4057.
[62] Zhou J Y, Shi J, Li G. Fine tuning support vector machines for short-term wind speedforecasting[J]. Energy conversion and management2011,52:1990-1998.
[63] Liu H, Tian H Q, Chen C, Li Y F. A hybrid statistical method to predict wind speed andwind power[J]. Renew energy,2010,35(8):1857-1861.
[64] Catalao J P S, Pousinho H M I, Mendes V M F. Short-term wind power forecasting inPortugal by neural networks and wavelet transform[J]. Renew energ2011,36(4):1245-1251.
[65] Girosi F. Models of noise and robust estimates. A. I. Memo1287, Artificial intelligencelaboratory, Massachusetts institute of technology,1991.
[66] Pontil M, Mukherjee S and Girosi F. On the noise model of support vector machinesregression. A. I. Memo1651, Artificial intelligence laboratory, Massachusetts instituteof technology,1998.
[67] Pontil M, Mukherjee S, Girosi F. On the noise model of support vector machinesregression [J]. ALT,2000, LNAI:316-324.
[68] Chu W, Keerthi S S, Ong C J. Bayesian support vector regression using a unified lossfunction [J]. IEEE Trans.on neural networks,2004,22(1):29-44.
[69] Klaus-Robert M, Sebastian M. An introduction to kernel-based learning algorithms [J].IEEE Trans. on neural networks,2001,12(2):181-202.
[70] Huber P J. Robust estimation of a location parameter [J]. The annals of mathematicalstatistics,1964,35(1):73-101.
[71] Huber P J. Robust statistics [M]. New York: Wiley,1981.
[72] Fan S, Liao J R, et al. Forecasting the wind generation using a two-stage network basedon meteorological information [J]. IEEE Trans. on energy conversion,2009,24(2):474-482.
[73] Guo Z H, Zhao J, Zhang W Y, Wang J Z. A corrected hybrid approach for wind speedprediction in Hexi Corridor of China [J]. Energy,2011,36:1668-1679.
[74]邓乃杨,田英杰.数据挖掘中的新方法:支持向量机[M].科学出版社,2004,6.
[75]邓乃杨,田英杰.支持向量机:理论、算法与拓展[M].科学出版社,2009,8.
[76]田英杰.支持向量回归机及其应用研究[D].北京:中国农业大学,2005,6.
[77] Rockafellar R T. The multiplier method of hestenes and powell applied to convexprogramming[J]. Journal of optimization theory and applixations,1973,12(6):555-562.
[78] Rockafellar R T. Augmented Lagrange multiplier functions and duality in non-convexprogramming[J]. SIAM Journal on control,1974,12(2),268-285.
[79] Boyd S, Vandenberghe L. Convex optimization[M]. Cambridge University Press,2004.
[80]马昌凤.最优化方法及其Matlab程序设计.科学出版社,2010.8.
[81]杨志民,刘广利.不确定性支持向量机原理及应用[M].科学出版社,2007,4.
[82]杨志民,刘广利.不确定性支持向量机:算法及应用[M].科学出版社,2012,1.
[83] Bottou L. Stochastic gradient descent on toy problems,2007. http://leon.bottou.org/projects/sgd.
[84] Bottou L. Large-scale machine learning with stochastic gradient descent[C]. Proceed-ings of the19th international conference on computational statistics(COMPSTAT'2010),Paris, Springer, August2010,177–187.
[85] Bordes A, Bottou L, Gallinari P. SGD-QN: careful quasi Newton stochastic gradientdescent. Journal of machine learning research,2009,10:1737-1754.
[86] Wang Z, Crammer K, Vucetic W. Breaking the curse of kernelization: budgetedstochastic gradient descent for large-scale SVM training. Journal of machine learningresearch,2012,13:3103-3131.
[87] Pawlak Z. Rough sets [J]. International Journal of computer and information science.1982,11(5):341-356.
[88] Pawlak Z. Rough sets: theoretical aspects of reasoning of data [M]. Dordrecht: KluwerAcademic Publishers,1991.
[89] Pawlak Z, Skoworn A. Rudiments of rough sets [J]. Information science,2007,177:3-27.
[90] Bonikowaski Z., Algebraic structrues of rough sets. Fuzzy sets and knowledge discovery,Springer-Verlag, Berlin,1995:242-247.
[91] Yao Y Y. Constructive and algebraic methods of the theory of rough sets. Informationsciences,1998,109:21-47.
[92] Bonikowaski Z. Algebraic structrues of rough sets. Fuzzy sets and knowledge discovery,Springer-Verlag, Berlin,1995:242-247.
[93] Liu G L, Zhu W. The algebraic structures of generalized rough set theory. Informationsciences,2008,178:4105–4113.
[94] Zhang W X, Mi J S, Wu W Z. Approaches to knowledge reductions in inconsistentsystems. International journal of intelligent systems,2003,18:989-1000.
[95]张文修,梁怡,吴伟志.信息系统与知识发现[M].北京:科学出版社,2003,9.
[96]胡清华,于达仁.应用粗糙计算[M].北京:科学出版社,2012,6.
[97]王国胤,姚一豫,于洪.粗糙集理论与应用研究综述[J].计算机学报,2009,32(7):1229-1247.
[98] Mi J S, Wu W Z, Zhang W X. Approaches to knowledge reductions based on variableprecision rough sets model. Information sciences,2004,159:255-272.
[99] Zhao Y P, Sun J G. Rough-support vector regression [J]. Expert systems withapplications [J].2009,36(6):9793–9798.
[100] Lingras P, Butz C J. Rough support vector regression [J]. European journal ofoperational research.2010,206(2):445–455.
[101]高爽,冬雷,高阳,廖晓钟.基于粗糙集理论的中长期风速预测[J].中国电机工程学报,2012,32(1):32-37.
[102]韩虎.基于粗糙集理论的中长期风速预测[D].兰州：兰州交通大学,2011,12.
[103] Yang H, Chan L, King I. Support vector machine regression for volatile stockmarketprediction [C]. In: Yin, H, Allinson, N, Freeman, R, Keane, J, Hubbard, S (Eds.),Intelligent data engineering and automated learning. Springer, LNCS,2002,2412:391-396.
[104] Yang H. Margin variations in support vector regression for the stock market prediction.M. Phil. dissertation, department of computer science and engineering, The ChineseUniversity of Hong Kong. http://www.svms.o rg/finance/Yang2003.
[105] Fiacco, A.V. and McCormick, G.P. Nonlinear programming. Sequential unconstrainedminimization techniques. Society or industrial and applied mathematics. First publishedin1968by research analysis corporation.
[106] Chen, D.S. and Jian, R.C. A robust backpropagation learning algorithm for function ap-proximation. IEEE Trans. on neural networks,1994,5(3):467-479.