遗传算法在电力市场交易决策中的应用研究
|
摘要
电力市场是指发电商、输电商、售电商和电力用户就电力及相关服务按市场规则进行公平交易的场所。电力市场交易决策是电力市场运营的核心，其功能在于根据发电商的报价曲线，在满足机组、电网和市场交易等约束条件下，对发电商投标的发电量和电价做出决策，使得电网总购电费用最小。电力市场的特殊性决定了电力市场交易决策的特殊性。如何用数学模型描述电力市场交易决策，通过何种算法求解该决策模型，如何高效求解等问题对我国正在建立的电力市场具有指导意义，是电力市场研究领域的前沿课题，具有重大的理论价值和应用前景。
本文通过对传统电力调度模式和电力市场交易模式的比较分析，建立了电力市场交易决策的数学模型。该数学模型考虑了机组组合问题，即机组启停费用问题，同时还考虑了以边际电价和机组实际报价结算的两种结算方式。对于这一大规模、非线性决策问题，传统优化方法或多或少存在不足。遗传算法是建立在自然选择和遗传机理基础上的迭代自适应概率性搜索算法，具有全局优化、鲁棒性和并行性等特点，能够弥补传统优化方法的不足。因此，本文提出基于遗传算法求解电力市场交易决策数学模型。针对标准遗传算法中存在的问题，提出了一个十进制-二进制混合编码的遗传算法，并结合电力市场交易决策模型的特殊性，对算法设计和求解中的若干关键技术，包括：编码方法、约束处理技术、遗传算子设计和算法效率优化等进行了研究。为提高算法效率及有利于并行计算，本文采用面向对象的技术设计和实现遗传算法，运用统一建模语言对遗传算法进行面向对象建模，描述了种群和个体的静态特性和动态特性，给出了类图、交互图和类中的方法。
运用本文提出的算法和程序，对线性报价曲线的电力市场交易决策进行了实例求解，获得了有实际意义的结果，表明基于遗传算法求解电力市场交易决策模型的有效性。
Electric power market is a market where electric power and other auxiliary services are exchanged fairly according to the market rules between generation suppliers, transmission service suppliers and demand service suppliers. Electric power market transaction decision-making is a core of electric power market operation and its aim is to make transmission service supplier get maximum profit through competition between generation suppliers under constrains of unit, grid and market. Specialty of electric power market leads to specialty of electric power market transaction decision-making. Some problems such as how to describe electric power market transaction decision-making using mathematic model, how to solve this model using algorithms and how to effectively solve it are very important theoretically and practically and will be useful to the construction of electric power market in China.
Under comparative analysis of traditional power dispatching model and electric power market transaction model, an electricity market transaction decision-making mathematic model is constructed. This model considers unit commitment or unit fees of startup and suspension, and it also considers two clearing methods based on unified marginal cost and practical bidding price. Aimed at defects of traditional optimization methods, GA(Genetic Algorithm) is proposed to solve electric power market transaction decision-making mathematic model. Based on natural selection and genetic mechanism, GA is an algorithm with characteristics of global optimization and parallelism. It can make up defects of traditional optimization methods. So a GA based method is proposed to solve this model. Because of some defects in standard Genetic Algorithms based on binary encoding and decoding techniques, a decimal-binary hybrid coding GA is proposed and according to the specialty of the model, some critical techniques, such as coding met
hod, constrains-handling technique, genetic operator design and algorithm efficiency optimization method, are studied. Genetic Algorithm programs based on object-oriented design and implementation are suggested to improve algorithm efficiency and to favor parallel computation. Unified Modeling Language is used to model object-oriented
Genetic Algorithm program. Class diagrams, interactive diagrams and methods of class
are given in the essay.
Using GA and its programs proposed in the essay, electric power market transaction decision-making mathematic model with linear bidding curve is analyzed and solved, and practically useful results are given. It proves that it is effective to solve electric power market transaction decision-making mathematic model through GA and its programs.
本文通过对传统电力调度模式和电力市场交易模式的比较分析，建立了电力市场交易决策的数学模型。该数学模型考虑了机组组合问题，即机组启停费用问题，同时还考虑了以边际电价和机组实际报价结算的两种结算方式。对于这一大规模、非线性决策问题，传统优化方法或多或少存在不足。遗传算法是建立在自然选择和遗传机理基础上的迭代自适应概率性搜索算法，具有全局优化、鲁棒性和并行性等特点，能够弥补传统优化方法的不足。因此，本文提出基于遗传算法求解电力市场交易决策数学模型。针对标准遗传算法中存在的问题，提出了一个十进制-二进制混合编码的遗传算法，并结合电力市场交易决策模型的特殊性，对算法设计和求解中的若干关键技术，包括：编码方法、约束处理技术、遗传算子设计和算法效率优化等进行了研究。为提高算法效率及有利于并行计算，本文采用面向对象的技术设计和实现遗传算法，运用统一建模语言对遗传算法进行面向对象建模，描述了种群和个体的静态特性和动态特性，给出了类图、交互图和类中的方法。
运用本文提出的算法和程序，对线性报价曲线的电力市场交易决策进行了实例求解，获得了有实际意义的结果，表明基于遗传算法求解电力市场交易决策模型的有效性。
Electric power market is a market where electric power and other auxiliary services are exchanged fairly according to the market rules between generation suppliers, transmission service suppliers and demand service suppliers. Electric power market transaction decision-making is a core of electric power market operation and its aim is to make transmission service supplier get maximum profit through competition between generation suppliers under constrains of unit, grid and market. Specialty of electric power market leads to specialty of electric power market transaction decision-making. Some problems such as how to describe electric power market transaction decision-making using mathematic model, how to solve this model using algorithms and how to effectively solve it are very important theoretically and practically and will be useful to the construction of electric power market in China.
Under comparative analysis of traditional power dispatching model and electric power market transaction model, an electricity market transaction decision-making mathematic model is constructed. This model considers unit commitment or unit fees of startup and suspension, and it also considers two clearing methods based on unified marginal cost and practical bidding price. Aimed at defects of traditional optimization methods, GA(Genetic Algorithm) is proposed to solve electric power market transaction decision-making mathematic model. Based on natural selection and genetic mechanism, GA is an algorithm with characteristics of global optimization and parallelism. It can make up defects of traditional optimization methods. So a GA based method is proposed to solve this model. Because of some defects in standard Genetic Algorithms based on binary encoding and decoding techniques, a decimal-binary hybrid coding GA is proposed and according to the specialty of the model, some critical techniques, such as coding met
hod, constrains-handling technique, genetic operator design and algorithm efficiency optimization method, are studied. Genetic Algorithm programs based on object-oriented design and implementation are suggested to improve algorithm efficiency and to favor parallel computation. Unified Modeling Language is used to model object-oriented
Genetic Algorithm program. Class diagrams, interactive diagrams and methods of class
are given in the essay.
Using GA and its programs proposed in the essay, electric power market transaction decision-making mathematic model with linear bidding curve is analyzed and solved, and practically useful results are given. It proves that it is effective to solve electric power market transaction decision-making mathematic model through GA and its programs.
引文
[1] 赵遵廉、辛耀中、郭国川．电力市场运营系统[M]．北京：中国电力出版社，2001．
[2] 于尔铿，周京阳，张学松．电力市场竞价模型与原理[J]．电力系统自动化，2001，25(1)：24-27．
[3] 于尔铿，周京阳，吴玉生．发电报价曲线研究[J]．电力系统自动化，2001，25(2)：23-26．
[4] 于尔铿，周京阳，吴玉生．发电竞价算法(一)——排队法[J]．电力系统自动化，2001，25(4)：16—19．
[5] 于尔铿，周京阳，吴玉生，王功涛．发电竞价算法(二)——等报价法[J]．电力系统自动化，2001，25(5)：11—13．
[6] 吴玉生，周京阳，王功涛，于尔铿．发电竞价算法(三)——动态规划法[J]．电力系统自动化，2001，25(6)：19—22，65．
[7] 周京阳，吴玉生，王功涛，于尔铿．发电竞价算法(四)——网络流规划法[J]．电力系统自动化，2001，25(7)：23—27．
[8] 王功涛，于尔铿，周京阳，吴玉生．发电竞价算法(五)——线性规划法[J]．电力系统自动化，2001，25(8)：20—23，33．
[9] 汪峰，白晓民．面向电力市场交易计划算法研究[J]．电网技术，2001，25(8)：45—49．
[10] Holland J H. Adaption in natural and artificial systems. Boston: MIT Press, 1992.
[11] Michalewicz Z.Genetic Algorithms+Data Structures =Evolution on Programs. Berlin: Springer-Verlag, 1996.
[12] Fogel D B. An Introduction to Simulated Evolutionary Optimization. IEEE Trans on Neural Network, 1994, 5(1): 3-14.
[13] Fogel D B. Evolutionary Computations: Toward a New Philosophy of Machine Intelligence. New York: IEEE Press, 1995.
[14] Goldberg D E. Genetic Algorithms in Search, Optimization and Machine Learning. Addison-Wesley Publishing Company, 1989.
[15] 曹一家，程时杰．进化算法在工程应用中的若干实用技术[J]．电力系统自动化，2001，25(1)：62—65；2001，25(2)：60—63．
[16] 文福拴，韩祯祥．模拟进化优化方法在电力系统中的应用综述[J]．电力系统自动化，1996，
20(1): 59-62; 1996, 20(2): 60-62; 1996, 20(3): 72-74.
[17] Grefenstette J J. Optimization of Control Parameters for Genetic Algorithms. IEEE Trans on System, Man and Cybernetics, 1986, 16(1): 122-128.
[18] C.W. Richter, G.B. Sheble. Genetic Algorithm Evolution of Utility Bidding Strategies for the Competitive Marketplace. IEEE Trans on Power Systems, 1998, 13(1): 256-161.
[19] C.W. Richter, G.B. Sheble, D. Ashlock. Comprehensive Bidding Strategies with Genetic Programming/Finite State Automata. IEEE Trans on Power Systems, 1999, 14(4): 1207-1212.
[20] C.W. Richter, G.B. Sheble. A Profit-Based Unit Commitment GA for the Comprehensive Environment. IEEE Trans on Power Systems, 2000, 15(12): 715-721.
[21] T. Alvy, D. Goodwin, X.M.Wang. A Security-Constrained Bid-Clearing System for the New Zealand and Wholesale Electricity Market. IEEE Trans on Power Systems, 1998, 13(2): 340-346.
[22] K. W. Cheung, D.Sun. Energy and Ancillary Service Dispatch for the Interim ISO New England Electricity Market. IEEE Trans on Power Systems, 2000, 15(3): 968-974.
[23] X. Ma, D. Sun, K.W.Cheung. Energy and Reserve Dispatch in a Multi-Zone Electricity Market. IEEE Trans on Power Systems, 1999, 14(3): 913-919.
[24] 韩祯祥，文福拴．模拟进化优化方法及其应用——遗传算法[J]．计算机科学，1995，22(2)：47—56．
[25] 文福拴，A．K．David．电力市场中的投标策略[J]．电力系统自动化，2000，24(14)：1—6．
[26] F. C. Schweppe, R.D. Tabors, J. L. Kirtley. Homeostatic Utility Control. IEEE Trans on Power Apparatus and Systems, 1980, PAS-99(3): 1151-1163.
[27] M. C. Caramanis, R. E. Bohn, F.C. Schweppe. Optimal Spot Pricing: Practice and Theory. IEEE Trans on Power Apparatus and Systems, 1982, PAS-101(3): 3234-3245.
[28] 潘正君、康立山、陈毓屏．演化计算[M]．北京：清华大学出版社，南宁：广西科学技术出版社，1998．
[29] 陈国良、王煦法、庄镇泉、王东生．遗传算法及其应用[M]．北京：人民邮电出版社，1996．
[30] 王正志、薄涛．进化计算[M]．长沙：国防科技大学出版社，2001．
[31] 印曼．Java语言与面向对象程序设计[M]．北京：清华大学出版社，2000．
[32] 蔡自兴、徐光佑．人工智能及其应用[M]．北京：清华大学出版社，1996．
[33] 康立山、谢云、尤矢勇、罗祖华．非数值并行算法(第一册)：模拟退火算法[M]．北京：科学出版社，1994．
[34] 刘勇、康立山、陈毓屏．非数值并行算法(第二册)：遗传算法[M]．北京：科学出版社，1995．
[35] Li Yuanxiang and Kang Lishan. Lattice models for first order wave equations, in S.K. Aityan etc. (Eds). Proceedings of Neural, Parallel and Scientific Computations, Dynamic Publishers Inc, U.S.A., (1995): 489-496.
[36] Xiong Shengwu and Li Yuanxiang. Evolutionary algorithms for solving parameter identification problem of PDEs. Advances in Computer Science and Technology, 1999, 1: 55-19.
[37] Li Yuanxiang and Xiong Shengwu. A class of Lattice Boltzmann models with Energy Equations. Lecture Notes in Physics, Vol. 552, 2000, ⅩⅪ, Vlarge Spinger, 208-215.
[38] Xu Jinwen and Li Yuanxiang. The Parallel Calculation of the Ground-state correlation Energy Of Helium Atom. Wuhan University Journal of Natural Sciences, Vol. 4, No.1, 1999, 45-48.
[39] Li Yuanxiang and Xu Jinwen. Using Evolutionary Algorithm to Calculate the Ground-state Energy of Double-electron Atoms in a Uniform Magnetic Field. International Journal of Modern Physics C, Vol. 11, No.1, 2000, 183-194.
[40] Zhang Jinbo and Li Yuanxiang. A Practical Evolutionary Algorithm of One Dimensional Stock Cutting Problem, Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun28?July2, 2000, Hefei, P. R. China, pp. 3729, IEEE Networking the world。
[41] Li Jun and Li Yuanxiang. Parallel Training Algorithm of BP Neural Networks. Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun28-July2, 2000, Hefei, P. R. China, pp. 872, IEEE Networking the world.
[42] Li Yuanxiang and Xie Man. An Improvement to the Programming of EHW Design, System Engineering, System Science and Complexity Research, Ed. Jifa Gu, Research InformationLtd., 2000, pp. 662.
[43] Yang Yiwei and Li Yuanxiang. Designing Evolvable Hardware (EHW) Based on
Combinational Logical Circuits. Proceedings of Technologies and Applications of System simulations' 2000, pp. 445.
[44] 杨波．发电侧电力市场交易决策系统的分析与设计[J]．华中电力，2003，16(1)：8—11．
[45] 杨波．基于UML的电力物资管理系统的建模与开发[J]．华中电力，2003，16(3)：67—70．
[46] 杨波．基于PKI的电力市场安全交易模型及实现方法[J]．电力信息化，2003，(3)：61—63．
[47] 杨波、高飞．基于构件的电力市场运营平台的研究[J]．华中电力，2004，17(2)：7—10．
[2] 于尔铿，周京阳，张学松．电力市场竞价模型与原理[J]．电力系统自动化，2001，25(1)：24-27．
[3] 于尔铿，周京阳，吴玉生．发电报价曲线研究[J]．电力系统自动化，2001，25(2)：23-26．
[4] 于尔铿，周京阳，吴玉生．发电竞价算法(一)——排队法[J]．电力系统自动化，2001，25(4)：16—19．
[5] 于尔铿，周京阳，吴玉生，王功涛．发电竞价算法(二)——等报价法[J]．电力系统自动化，2001，25(5)：11—13．
[6] 吴玉生，周京阳，王功涛，于尔铿．发电竞价算法(三)——动态规划法[J]．电力系统自动化，2001，25(6)：19—22，65．
[7] 周京阳，吴玉生，王功涛，于尔铿．发电竞价算法(四)——网络流规划法[J]．电力系统自动化，2001，25(7)：23—27．
[8] 王功涛，于尔铿，周京阳，吴玉生．发电竞价算法(五)——线性规划法[J]．电力系统自动化，2001，25(8)：20—23，33．
[9] 汪峰，白晓民．面向电力市场交易计划算法研究[J]．电网技术，2001，25(8)：45—49．
[10] Holland J H. Adaption in natural and artificial systems. Boston: MIT Press, 1992.
[11] Michalewicz Z.Genetic Algorithms+Data Structures =Evolution on Programs. Berlin: Springer-Verlag, 1996.
[12] Fogel D B. An Introduction to Simulated Evolutionary Optimization. IEEE Trans on Neural Network, 1994, 5(1): 3-14.
[13] Fogel D B. Evolutionary Computations: Toward a New Philosophy of Machine Intelligence. New York: IEEE Press, 1995.
[14] Goldberg D E. Genetic Algorithms in Search, Optimization and Machine Learning. Addison-Wesley Publishing Company, 1989.
[15] 曹一家，程时杰．进化算法在工程应用中的若干实用技术[J]．电力系统自动化，2001，25(1)：62—65；2001，25(2)：60—63．
[16] 文福拴，韩祯祥．模拟进化优化方法在电力系统中的应用综述[J]．电力系统自动化，1996，
20(1): 59-62; 1996, 20(2): 60-62; 1996, 20(3): 72-74.
[17] Grefenstette J J. Optimization of Control Parameters for Genetic Algorithms. IEEE Trans on System, Man and Cybernetics, 1986, 16(1): 122-128.
[18] C.W. Richter, G.B. Sheble. Genetic Algorithm Evolution of Utility Bidding Strategies for the Competitive Marketplace. IEEE Trans on Power Systems, 1998, 13(1): 256-161.
[19] C.W. Richter, G.B. Sheble, D. Ashlock. Comprehensive Bidding Strategies with Genetic Programming/Finite State Automata. IEEE Trans on Power Systems, 1999, 14(4): 1207-1212.
[20] C.W. Richter, G.B. Sheble. A Profit-Based Unit Commitment GA for the Comprehensive Environment. IEEE Trans on Power Systems, 2000, 15(12): 715-721.
[21] T. Alvy, D. Goodwin, X.M.Wang. A Security-Constrained Bid-Clearing System for the New Zealand and Wholesale Electricity Market. IEEE Trans on Power Systems, 1998, 13(2): 340-346.
[22] K. W. Cheung, D.Sun. Energy and Ancillary Service Dispatch for the Interim ISO New England Electricity Market. IEEE Trans on Power Systems, 2000, 15(3): 968-974.
[23] X. Ma, D. Sun, K.W.Cheung. Energy and Reserve Dispatch in a Multi-Zone Electricity Market. IEEE Trans on Power Systems, 1999, 14(3): 913-919.
[24] 韩祯祥，文福拴．模拟进化优化方法及其应用——遗传算法[J]．计算机科学，1995，22(2)：47—56．
[25] 文福拴，A．K．David．电力市场中的投标策略[J]．电力系统自动化，2000，24(14)：1—6．
[26] F. C. Schweppe, R.D. Tabors, J. L. Kirtley. Homeostatic Utility Control. IEEE Trans on Power Apparatus and Systems, 1980, PAS-99(3): 1151-1163.
[27] M. C. Caramanis, R. E. Bohn, F.C. Schweppe. Optimal Spot Pricing: Practice and Theory. IEEE Trans on Power Apparatus and Systems, 1982, PAS-101(3): 3234-3245.
[28] 潘正君、康立山、陈毓屏．演化计算[M]．北京：清华大学出版社，南宁：广西科学技术出版社，1998．
[29] 陈国良、王煦法、庄镇泉、王东生．遗传算法及其应用[M]．北京：人民邮电出版社，1996．
[30] 王正志、薄涛．进化计算[M]．长沙：国防科技大学出版社，2001．
[31] 印曼．Java语言与面向对象程序设计[M]．北京：清华大学出版社，2000．
[32] 蔡自兴、徐光佑．人工智能及其应用[M]．北京：清华大学出版社，1996．
[33] 康立山、谢云、尤矢勇、罗祖华．非数值并行算法(第一册)：模拟退火算法[M]．北京：科学出版社，1994．
[34] 刘勇、康立山、陈毓屏．非数值并行算法(第二册)：遗传算法[M]．北京：科学出版社，1995．
[35] Li Yuanxiang and Kang Lishan. Lattice models for first order wave equations, in S.K. Aityan etc. (Eds). Proceedings of Neural, Parallel and Scientific Computations, Dynamic Publishers Inc, U.S.A., (1995): 489-496.
[36] Xiong Shengwu and Li Yuanxiang. Evolutionary algorithms for solving parameter identification problem of PDEs. Advances in Computer Science and Technology, 1999, 1: 55-19.
[37] Li Yuanxiang and Xiong Shengwu. A class of Lattice Boltzmann models with Energy Equations. Lecture Notes in Physics, Vol. 552, 2000, ⅩⅪ, Vlarge Spinger, 208-215.
[38] Xu Jinwen and Li Yuanxiang. The Parallel Calculation of the Ground-state correlation Energy Of Helium Atom. Wuhan University Journal of Natural Sciences, Vol. 4, No.1, 1999, 45-48.
[39] Li Yuanxiang and Xu Jinwen. Using Evolutionary Algorithm to Calculate the Ground-state Energy of Double-electron Atoms in a Uniform Magnetic Field. International Journal of Modern Physics C, Vol. 11, No.1, 2000, 183-194.
[40] Zhang Jinbo and Li Yuanxiang. A Practical Evolutionary Algorithm of One Dimensional Stock Cutting Problem, Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun28?July2, 2000, Hefei, P. R. China, pp. 3729, IEEE Networking the world。
[41] Li Jun and Li Yuanxiang. Parallel Training Algorithm of BP Neural Networks. Proceedings of the 3rd World Congress on Intelligent Control and Automation, Jun28-July2, 2000, Hefei, P. R. China, pp. 872, IEEE Networking the world.
[42] Li Yuanxiang and Xie Man. An Improvement to the Programming of EHW Design, System Engineering, System Science and Complexity Research, Ed. Jifa Gu, Research InformationLtd., 2000, pp. 662.
[43] Yang Yiwei and Li Yuanxiang. Designing Evolvable Hardware (EHW) Based on
Combinational Logical Circuits. Proceedings of Technologies and Applications of System simulations' 2000, pp. 445.
[44] 杨波．发电侧电力市场交易决策系统的分析与设计[J]．华中电力，2003，16(1)：8—11．
[45] 杨波．基于UML的电力物资管理系统的建模与开发[J]．华中电力，2003，16(3)：67—70．
[46] 杨波．基于PKI的电力市场安全交易模型及实现方法[J]．电力信息化，2003，(3)：61—63．
[47] 杨波、高飞．基于构件的电力市场运营平台的研究[J]．华中电力，2004，17(2)：7—10．