基于谱质心迁移的线损优化分摊方法研究
|
摘要
线损优化与分摊是我国电力市场改革中亟待解决的技术问题。提出一种基于谱质心迁移的线损优化分摊方法,以应对工业环境与模拟环境下线损优化分摊模型的领域适应性问题。首先计算各源领域与目标领域之间的电频谱相似度量(PFSSM),并据此将与目标领域最接近的源领域选为初始训练集,其次采用谱质心迁移模型剔除距离训练集谱质心均值较远的样本,并加入同数量的目标领域无标签样本,直至训练集谱质心均值与目标领域一致。最后基于我国某地区电力市场实际数据构造算例,试验结果表明通过计算源领域与目标领域之间的电频谱相似度量值,负载控制对线损优化分摊的影响小于线损分摊计算法则和母线运行方式。
Line loss optimization and allocation is a technical problem that needs to be solved urgently in China's power market reform.A spectral centroid transfer-based line loss optimization allocation strategy is proposed to adopt differentiated working conditions, such as real industrial and simulated environment, to solve the problem of field adaptability for line loss optimization allocation model.In this model, the initial training set is selected according to the power frequency spectrum similarity measure(PFSSM) values, which are calculated between different source domains and target domain.Then, some samples whose spectral centroid keeps away from the center of training set are removed, meanwhile the same quantity of samples from target domain are added.The transfer process finishes until the spectral centroid mean of training set is fit with the target domain. Experiments using an example of the actual data structure of the electricity market in a certain region was carried out, which solve the problem of the field adaptation for migration learning. By calculating similar electric spectrum similarity between the source domain and the target domain, the impact of load control on the line loss optimization allocation is less than the line loss allocation calculation rule and the bus operation mode.
Line loss optimization and allocation is a technical problem that needs to be solved urgently in China's power market reform.A spectral centroid transfer-based line loss optimization allocation strategy is proposed to adopt differentiated working conditions, such as real industrial and simulated environment, to solve the problem of field adaptability for line loss optimization allocation model.In this model, the initial training set is selected according to the power frequency spectrum similarity measure(PFSSM) values, which are calculated between different source domains and target domain.Then, some samples whose spectral centroid keeps away from the center of training set are removed, meanwhile the same quantity of samples from target domain are added.The transfer process finishes until the spectral centroid mean of training set is fit with the target domain. Experiments using an example of the actual data structure of the electricity market in a certain region was carried out, which solve the problem of the field adaptation for migration learning. By calculating similar electric spectrum similarity between the source domain and the target domain, the impact of load control on the line loss optimization allocation is less than the line loss allocation calculation rule and the bus operation mode.
引文
[1]祖凯涛.林卓琼.配电网精细化线损计算及降损分析[J].陕西电力,2016,44(1):60-64.ZU Kaitao, LIN Zhuoqiong. Line loss fine calculation and loss-reducing analysis[J].Shaanxi Electric Power,2016,44(1):60-64.
[2]应丽云,刘敏,邓磊,等.配电网降损综述[J].电力系统保护与控制,2017,45(19):162-169.YING Liyun, LIU Min, DENG Lei, et al. A comprehensive review of the loss reduction in distribution network[J].Power System Protection and Control,2017,45(19):162-169.
[3]王若丞.配电网三相不平衡对线损增加率及电压偏移的影响[J].电力工程技术,2017,36(4):131-135.WANG Ruocheng. The influence of unbalanced threephase on line loss and voltage offset in low voltage distribution network[J]. Electric Power Engineering Technology,2017,36(4):131-135.
[4]王义贺,含分布式发电的配电网网损分摊方法研究[D].北京:华北电力大学,2013.
[5] ISLAM M M, KIM J, KHAN S A, et al.Reliable bearing fault diagnosis using Bayesian inference-based multiclass support vector machines[J].Journal of the Acoustical Society of America, 2017, 141(2):89-92.
[6] PAHON E, STEINER N Y, JEMEI S, et al. Solid oxide fuel cell fault diagnosis and ageing estimation based on wavelet transform approach[J]. International Journal of Hydrogen Energy, 2016,41(31):13678-13687.
[7]李卫东,孙辉,武亚光.潮流追踪迭代算法[J].中国电机工程学报,2001,21(11):1-5.LI Weidong, SUN Hui, WU Yaguang. Power flow tracking iterative algorithms[J]. Proceedings of the CSEE,2001,21(11):1-5.
[8] BIALEK J.Topological generation and load distribution factors for supplementary charge allocation in transmission open access[J]. IEEE Transactions on Power Systems,1997,12(3):1185-1193.
[9] WANG Yifeng, GUO Chuangxin, WANG Yue. An analytical algorithm of complex power tracing based on equivalent admittance matrix operation[J]. Proceedings of the CSEE, 2015,35(5):1127-1134.
[10] YANG L, ZHANG J.Automatic transfer learning for short text mining[J]. Eurasip Journal on Wireless Communications&Networking, 2017,(1):42.
[11] ZHOU G, ZENG Z, HUANG J X, et al.Transfer learning for cross-lingual sentiment classification with weakly shared deep neural networks[C]. Pisa:International Acm Sigir Conference on Research&Development in Information Retrieval, ACM, 2016:245-254.
[12] GOUSSIES N A, UBALDE S, ME J AIL M. Transfer learning decision forests for gesture recognition[J].Journal of Machine Learning Research, 2017, 15(1):3667-3690.
[13]姜礼平,李冬雷,季傲.基于兰氏距离的改进冲突证据合成方法[J].海军工程大学学报,2016, 28(6):1-4.JIANG Liping, LI Donglei, JI Ao. An improved conflict evidence synthetic method based on lance distance[J].Journal of Naval University of Engineering,2016,28(6):1-4.
[14]曾金芳,滕召胜.谱质心在电力谐波分析中的应用[J].中国电机工程学报,2013, 33(31):73-80.ZENG Jinfang, TENG Zhaosheng. Application of spectral centroid in power harmonic analysis[J]. Proceedings of the CSEE,2013,33(31):73-80.
[15] GAJIC B, PALIWAL K K. Robust speech recognition in noisy environments based on subband spectral centroid histograms[J]. IEEE Transactions on Audio Speech&Language Processing, 2006, 14(2):600-608.
[16]董永峰,张娜.改进谱质心算法在噪声源识别中的应用[J].大科技,2017,1(29):309-310.DONG Yongfeng, ZHANG Na. Application of improved spectral centroid algorithm in noise source identification[J].Super Science,2017,1(29):309-310.
[2]应丽云,刘敏,邓磊,等.配电网降损综述[J].电力系统保护与控制,2017,45(19):162-169.YING Liyun, LIU Min, DENG Lei, et al. A comprehensive review of the loss reduction in distribution network[J].Power System Protection and Control,2017,45(19):162-169.
[3]王若丞.配电网三相不平衡对线损增加率及电压偏移的影响[J].电力工程技术,2017,36(4):131-135.WANG Ruocheng. The influence of unbalanced threephase on line loss and voltage offset in low voltage distribution network[J]. Electric Power Engineering Technology,2017,36(4):131-135.
[4]王义贺,含分布式发电的配电网网损分摊方法研究[D].北京:华北电力大学,2013.
[5] ISLAM M M, KIM J, KHAN S A, et al.Reliable bearing fault diagnosis using Bayesian inference-based multiclass support vector machines[J].Journal of the Acoustical Society of America, 2017, 141(2):89-92.
[6] PAHON E, STEINER N Y, JEMEI S, et al. Solid oxide fuel cell fault diagnosis and ageing estimation based on wavelet transform approach[J]. International Journal of Hydrogen Energy, 2016,41(31):13678-13687.
[7]李卫东,孙辉,武亚光.潮流追踪迭代算法[J].中国电机工程学报,2001,21(11):1-5.LI Weidong, SUN Hui, WU Yaguang. Power flow tracking iterative algorithms[J]. Proceedings of the CSEE,2001,21(11):1-5.
[8] BIALEK J.Topological generation and load distribution factors for supplementary charge allocation in transmission open access[J]. IEEE Transactions on Power Systems,1997,12(3):1185-1193.
[9] WANG Yifeng, GUO Chuangxin, WANG Yue. An analytical algorithm of complex power tracing based on equivalent admittance matrix operation[J]. Proceedings of the CSEE, 2015,35(5):1127-1134.
[10] YANG L, ZHANG J.Automatic transfer learning for short text mining[J]. Eurasip Journal on Wireless Communications&Networking, 2017,(1):42.
[11] ZHOU G, ZENG Z, HUANG J X, et al.Transfer learning for cross-lingual sentiment classification with weakly shared deep neural networks[C]. Pisa:International Acm Sigir Conference on Research&Development in Information Retrieval, ACM, 2016:245-254.
[12] GOUSSIES N A, UBALDE S, ME J AIL M. Transfer learning decision forests for gesture recognition[J].Journal of Machine Learning Research, 2017, 15(1):3667-3690.
[13]姜礼平,李冬雷,季傲.基于兰氏距离的改进冲突证据合成方法[J].海军工程大学学报,2016, 28(6):1-4.JIANG Liping, LI Donglei, JI Ao. An improved conflict evidence synthetic method based on lance distance[J].Journal of Naval University of Engineering,2016,28(6):1-4.
[14]曾金芳,滕召胜.谱质心在电力谐波分析中的应用[J].中国电机工程学报,2013, 33(31):73-80.ZENG Jinfang, TENG Zhaosheng. Application of spectral centroid in power harmonic analysis[J]. Proceedings of the CSEE,2013,33(31):73-80.
[15] GAJIC B, PALIWAL K K. Robust speech recognition in noisy environments based on subband spectral centroid histograms[J]. IEEE Transactions on Audio Speech&Language Processing, 2006, 14(2):600-608.
[16]董永峰,张娜.改进谱质心算法在噪声源识别中的应用[J].大科技,2017,1(29):309-310.DONG Yongfeng, ZHANG Na. Application of improved spectral centroid algorithm in noise source identification[J].Super Science,2017,1(29):309-310.