Suzhou Electric Appliance Research Institute
期刊號： CN32-1800/TM| ISSN1007-3175

SUBSCRIPTION MANAGEMENT

發(fā)行征訂

首頁 >> 發(fā)行征訂 >> 征訂方式

基于圖卷積神經(jīng)網(wǎng)絡(luò)的機(jī)組組合問題加速求解方法

來源：電工電氣發(fā)布時間：2024-04-07 09:07瀏覽次數(shù)：302

基于圖卷積神經(jīng)網(wǎng)絡(luò)的機(jī)組組合問題加速求解方法

曾貴華，劉明波
(華南理工大學(xué) 電力學(xué)院，廣東 廣州 510640)
 
    摘 要：針對傳統(tǒng)的精確優(yōu)化算法求解規(guī)模較大的機(jī)組組合問題面臨時間可行性的挑戰(zhàn), 提出了一種基于圖卷積神經(jīng)網(wǎng)絡(luò)的機(jī)組組合問題加速求解方法。將機(jī)組組合問題構(gòu)建為一個混合整數(shù)線性規(guī)劃模型，根據(jù)分支定界法的求解原理，將分支策略定義為從候選變量的特征到候選變量得分的映射關(guān)系；提出在離線階段使用圖卷積神經(jīng)網(wǎng)絡(luò)來模擬強(qiáng)分支策略的決策行為，并將學(xué)習(xí)到的映射關(guān)系應(yīng)用到在線分支過程中，從而加速分支定界法求解機(jī)組組合問題。通過 IEEE 39 節(jié)點 10 機(jī)組和 IEEE 118 節(jié)點 54 機(jī)組系統(tǒng)的算例分析，驗證了所提方法的有效性。
    關(guān)鍵詞： 發(fā)電機(jī)；機(jī)組組合；分支定界法；分支策略；圖卷積神經(jīng)網(wǎng)絡(luò)
    中圖分類號：TM744     文獻(xiàn)標(biāo)識碼：A     文章編號：1007-3175(2024)03-0044-07
 
Acceleration Solving Method for Unit Commitment Problem Based on
Graph Convolution Neural Network
 
ZENG Gui-hua, LIU Ming-bo
（School of Electric Power Engineering, South China University of Technology, Guangzhou 510640, China）
 
    Abstract: To solve the challenge of time feasibility faced by traditional accurate optimization algorithms for solving large-scale Unit Commitment (UC) problems, this paper proposes an accelerated solution method for solving the UC problems based on graph convolution neural network. Firstly, the UC problem is constructed as a Mixed Integer Linear Programming (MILP) model. Next, according to the solution principle of the branch-and-bound method, we define the branching strategy as a mapping relationship from the features of candidate variables to the scores of candidate variables. Thus, we propose to mimic the decision-making behavior of strong branching strategy in the offline phase using Graph Convolutional Neural Network (GCNN) and apply the learned mapping relationship to the online branching process to accelerate the process of the branch and bound method to solve the UC problem. Finally, the effectiveness of the proposed method is verified by the analysis of IEEE 39-node 10-unit and IEEE 118-node 54-unit systems.
    Key words: generator; unit commitment; branch and bound method; branch strategy; graph convolution neural network
 
參考文獻(xiàn)
[1] XAVIER Á S, QIU F, AHMED S.Learning to solve large-scale security-constrained unit commitment problems[J].INFORMS Journal on Computing，2021，33(2) ：739-756.
[2] SHOULTS R R, CHANG S K, HELMICK S, et al.A practical approach to unit commitment, economic dispatch and savings allocation for multiple-area pool operation with import/export constraints[J].IEEE Transactions on Power Apparatus and Systems,1980，PAS-99(2) ：625-635.
[3] BURNS R M, GIBSON C A.Optimization of priority lists for a unit commitment program[C]//Proceeding IEEE Power Engineering Society Summer Meeting，1975，453-461.
[4] LEE F N . The application of commitment utilization factor (CUF) to thermal unit commitment[J].IEEE Transactions on Power Systems，1991，6(2) ：691-698.
[5] PANG C K, CHEN H C.Optimal short-term thermal unit commitment[J].IEEE Transactions on Power Apparatus and Systems，1976，95(4) ：1336-1346.
[6] PANG C K, SHEBLÉ G B, ALBUYEH F.Evaluation of dynamic programming based methods and multiple area representation for thermal unit commitments[J].IEEE Transactions on Power Apparatus and Systems，1981，PAS-100(3) ：1212-1218.
[7] COHEN A I, YOSHIMURA M.A branch-and-bound algorithm for unit commitment [J] . IEEE Transactions on Power Apparatus and Systems，1983，PAS-102(2) ：444-451.
[8] 謝國輝，張粒子，舒雋，等. 基于分層分枝定界算法的機(jī)組組合[J] . 電力自動化設(shè)備，2009，29(12) ：29-32.
[9] HABIBOLLAHZADEH H, BUBENKO J A.Application of decomposition techniques to short-term operation planning of hydrothermal power system[J].IEEE Transactions on Power Systems，1986，1(1)：41-47.
[10] SASAK H, WATANAB M, KUBOKAWA J, et al.A solution method of unit commitment by artificial neural networks[J].IEEE Transactions on Power Systems，1992，7(3) ：974-981.
[11] LIANGL R H, KANG F C.Thermal generating unit commitment using an extended mean field annealing neural network[J].IEE Proceedings-Generation, Transmission and Distribution，2000，147(3)：164-170.
[12] JUSTE K A, KITA H, TANAKA E, et al.An evolutionary programming solution to the unit commitment problem[J].IEEE Transactions on Power Systems，1999，14(4) ：1452-1459.
[13] KAZARLIS S A, BAKIRTZIS A G, PETRIDIS V.A genetic algorithm solution to the unit commitment problem[J].IEEE Transactions on Power Systems，1996，11(1) ：83-92.
[14] LIN X, HOU Z J, REN H, et al.Approximate mixedinteger programming solution with machine learning technique and linear programming relaxation[C]//2019 3rd International Conference on Smart Grid and Smart Cities(ICSGSC).IEEE，2019 ：101-107.
[15] 張麗華. 基于內(nèi)點—分支定界法的最優(yōu)機(jī)組投入研究[D].南寧：廣西大學(xué)，2006.
[16] LINDEROTH J T, SAVELSBERGH M W P.A computational study of search strategies for mixed integer programming[J].INFORMS Journal on Computing，1999，11(2) ：173-187.
[17] APPLEGATE D, BIXBY R, CHVÁTAL V, et al.Finding cuts in the TSP (A preliminary report)[M].New Jersey: Rutgers University, New Brunswick, USA，1995 ：95-105.
[18] HE H, DAUME Ⅲ H, EISNER J M.Learning to search in branch and bound algorithms[J].Advances in Neural Information Processing Systems，2014，27 ：3293-3301.
[19] GASSE M, CHÉTELAT D, FERRONI N, et al.Exact combinatorial optimization with graph convolutional neural networks[J].Advances in Neural Information Processing Systems，2019，32 ：15554-15566.