基于最小二乘支持向量机和遗传算法的氧化铝悬浮焙烧能耗估计建模

doi:10.11772/j.issn.1001-9081.2014.04.1217

计算机应用 ›› 2014, Vol. 34 ›› Issue (4): 1217-1221.DOI: 10.11772/j.issn.1001-9081.2014.04.1217

基于最小二乘支持向量机和遗传算法的氧化铝悬浮焙烧能耗估计建模

刘代飞¹,尹吉¹,丁凤其²

1. 长沙理工大学能源与动力工程学院,长沙 410114；
2. 中南大学冶金与环境学院,长沙 410083

收稿日期:2013-09-18 修回日期:2013-11-15 出版日期:2014-04-01 发布日期:2014-04-29
通讯作者: 刘代飞
作者简介:刘代飞(1977-),男,湖南临武人,讲师，博士，主要研究方向:复杂工业过程建模、仿真优化与控制；
尹吉(1989-),男,湖南永兴人,硕士研究生，主要研究方向:热工过程自控化；
丁凤其(1962-),男,湖南湘乡人,副教授，高级工程师，主要研究方向:铝电解过程控制、氧化铝生产与智能控制。
基金资助:
湖南省教育厅资助项目;国家自然科学基金资助项目

Energy consumption estimation modeling of aluminum hydroxide gas suspension calcinations based on least squares support vector machine and genetic algorithm

LIU Daifei¹,YI Ji¹,DING Fengqi²

1. School of Energy and Power Engineering, Changsha University of Science and Technology, Changsha Hunan 410114, China;
2. School of Metallurgry and Environment, Central South University, Changsha Hunan 410083, China

Received:2013-09-18 Revised:2013-11-15 Online:2014-04-01 Published:2014-04-29
Contact: LIU Daifei

摘要/Abstract

摘要：

针对氧化铝悬浮焙烧能耗信息表征和模型应用的实际需求，建立一种最小二乘支持向量机(LS-SVM)能耗估计模型。基于该类模型结合遗传算法(GA)提出一种模型参数优化和工业应用策略。采用灰关联分析确定模型的主输入为主炉温度、烟气含氧量、原料含水量；采用K折交叉验证优化样本数据；采用比较模型预测误差确定核函数为径向基函数(RBF)核。建立输入为能耗参数，输出为模型标志的支持向量机工况模型选择器。能耗估计模型的自学习与动态优化通过样本的更新和聚类实现，模型的选择和投运通过模型选择器依据工况状态实施切换。实验结果表明，建立的焙烧能耗估计模型和模型应用策略，能提高模型的泛化能力、增强模型的工况适应性，是一种有效的焙烧能耗参数估计和分析方法。

Abstract:

According to the requirement of energy consumption information representation and model application in aluminum hydroxide gas suspension calcinations process, a kind of energy consumption estimation model was established based on Least Squares Support Vector Machine (LS-SVM) method. By combining the energy consumption model with Genetic Algorithm (GA), a kind of parameters optimization and industry application strategy was presented. Input parameters of energy estimation model were analyzed through grey relational analysis method, and the main factors of input parameters consisted of main furnace temperature, oxygen content of exhaust gas and containing water of aluminum hydroxide. The sampled data of energy consumption parameters were regrouped and optimized through K-fold cross-validation method. By comparing prediction accuracy of energy consumption models with various kernel functions, Radial Basis Function (RBF) kernel function was adopted to express feature information of sampling data. A model switcher whose inputs were energy parameters and output was symbol parameter of energy estimation model was constructed by Support Vector Machines (SVM) method. Self-learning and dynamic optimization processes of energy estimation model were realized by sample data updating and clustering. Model selection and application were realized by using the model switcher according to various calcinations conditions. The experimental results show that the LS-SVM modeling and application strategy can improve the generalization capability and conditions adaptability of energy estimation model. The presented strategy of model application is a feasible method for energy parameter analysis and estimation.

中图分类号:

TK39
TP183

刘代飞尹吉丁凤其. 基于最小二乘支持向量机和遗传算法的氧化铝悬浮焙烧能耗估计建模[J]. 计算机应用, 2014, 34(4): 1217-1221.

LIU Daifei YI Ji DING Fengqi. Energy consumption estimation modeling of aluminum hydroxide gas suspension calcinations based on least squares support vector machine and genetic algorithm[J]. Journal of Computer Applications, 2014, 34(4): 1217-1221.

参考文献

［1］LIU D F, DING F Q, ZHANG H L. Numerical simulation of high temperature air combustion in aluminum hydroxide gas suspension calcinations ［J］. Transactions of Nonferrous Metals Society of China, 2009, 19(1): 259-266.
［2］WEI H. Relationship of process parameters and capacity and consumption of Al(OH)3 gas suspension calciner ［J］. Light Metals, 2003(1): 13-18.(卫璜.Al(OH)3气态悬浮焙烧工艺参数与产能、消耗的关系［J］.轻金属,2003(1):13-18.)
［3］LI J, LIU D F, DAI X R, et al.Prediction of Al(OH)3 fluidized roasting temperature based on wavelet neural network ［J］. Transactions of Nonferrous Metals Society of China, 2007, 17(5): 1052-1056.
［4］WANG K, YANG J, WANG L, et al.Review on generalization problem of artificial neural network ［J］. Application Research of Computers, 2008, 25(12): 3525-3530.(王恺，杨巨峰，王立，等.人工神经网络泛化问题研究综述［J］.计算机应用研究,2008,25(12):3525-3530.)
［5］ZHAO Y, HU W. Improve the generalization capability of artificial neural network ［J］. Journal of Nanjing University of Information Science and Technology: Natural Science Edition, 2011, 3(2): 164-167.(赵远东，胡为尧.人工神经网络泛化性能改进［J］.南京信息工程大学学报:自然科学版,2011,3(2):164-167.)
［6］VAPNIK V N，GOLOWIEH S E, SMOLA A. Support vector method for function approximation, regression estimation, and signal processing ［C］// Adavanees in Neural Information Processing Systems 1996. Cambridge: MIT Press, 1997: 281-287.
［7］VAPNIK V N. The nature of statistical learning theory ［M］. ZHANG X, translated. Beijing: Tsinghua University Press: 2000: 126-150.(VAPNIK V N.统计学习理论本质［M］.张学工,译.北京:清华大学出版社,2000:126-150.)
［8］DING S, QI B, TAN H. An overview on theory and algorithm of support vector machines ［J］. Journal of University of Electronic Science and Technology of China, 2011, 40(1): 2-10.(邓世飞，齐丙娟，谭红艳.支持向量机理论与算法研究综述［J］.电子科技大学学报,2011,40(1):2-10.)
［9］SUYKENS J A K, van GESTEL T, de BRABANTER J, et al.Least squares support vector machines ［M］. Singapore: World Scientific, 2002: 71-111
［10］ISMAIL S, SHABRI A, SAMSUDIN R. A hybrid model of Self-organizing Maps (SOM) and Least Square Support Vector Machine (LSSVM) for time-series forecasting ［J］. Expert Systems with Applications, 2011, 38(8): 10574-10578.
［11］THI N, LEE G B, WHEELER P, et al.GA-SVM based framework for time series forecasting ［C］// ICNC 2009: Proceedings of the 5th International Conference on Natural Computation. Washington, DC: IEEE Computer Society, 2009: 493-498.
［12］TEOH E J, XIANG C. Feature selection and classification via a GA-SVM hybrid ［C］// GEM 2008: Proceedings of the 2008 International Conference on Genetic and Evolutionary Methods. Athens, GA: CSREA Press, 2008: 83-89.
［13］LIU H B, JIAO Y B. Application of genetic algorithm-support vector machine (GA-SVM) for damage identification of bridge ［J］. International Journal of Computational Intelligence and Applications, 2011, 10(4): 383-397.
［14］LIU S, GUO T, DANG Y. The theory and application of grey system ［M］. Beijing: Science and Technology Press, 1999: 49-111.(刘思峰，郭天榜，党耀国.灰色系统理论及应用［M］.北京：科学技术出版社,1999:49-111.)
［15］DING C, WANG L. BP neural network application in turbine fault diagnosis base on the cross-validation ［J］. Electric Power Science and Engineering, 2008, 24(3): 31-34.(丁常富，王亮.基于交叉验证法的BP神经网络在汽轮机故障诊断中的应用［J］.电力科学与工程,2008,24(3):31-34.)
［16］MORENO-TORRES J G, SAEZ J A, HERRERA F. Study on the impact of partition induced dataset shift on k-fold cross-validation ［J］. IEEE Transactions on Neural Networks and Learning Systems, 2012, 23(8):1304-1312.
［17］BAYKAN N A, YILMAZ N. A mineral classification system withmultiple artificial neural network using k-fold cross validation ［J］. Mathematical and Computational Applications, 2011, 16 (1): 22-30.
［18］FENG G. Parameter optimizing for support vector machines classification ［J］. Computer Engineering and Applications， 2011， 47(3): 123-124.(奉国和.SVM分类核函数及参数选择比较［J］.计算机工程与应用,2011,47(3):123-124.)

基于最小二乘支持向量机和遗传算法的氧化铝悬浮焙烧能耗估计建模

Energy consumption estimation modeling of aluminum hydroxide gas suspension calcinations based on least squares support vector machine and genetic algorithm

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

编辑推荐

Metrics