考虑小样本统计的BP神经网络配电系统可靠性预测方法
|
摘要
传统服务于系统规划的可靠性分析方法,由于多基于逻辑推理或统计分析,需要以足量‘故障—停电’事件匀质样本为建模保障,在面对配电系统结构动态变化以及稀少数据环境时,难以对指标进行精确估计。在此背景下,提出一种考虑小样本统计的BP神经网络配电系统可靠性指标预测方法。为保证神经网络训练样本的充足性,并保留小样本自身的统计规律,该文提出并比较Bootstrap和核密度拉丁超立方采样2种小样本增广技术,基于扩充后的样本对具有相同结构的神经网络模型进行参数训练,利用所得的神经网络对可靠性指标进行预测的精度作为选择合适扩充技术与神经网络结合的依据。通过预测用户年均停电时间的算例分析表明,利用Bootstrap小样本扩充技术和BP神经网络相结合的方法在小样本统计条件下具有更高的预测精度。
In system planning, traditional reliability evaluation methods are mainly based on logical reasoning or statistical analytics which needs enough ‘fault-outage' observations data samples for modeling. Therefore, reliability indices cannot be accurately estimated with dynamic changes of distribution system structure and insufficient data records. This paper proposed a BP-neural-network-based reliability forecasting method by considering the statistics of small sample. Firstly, in order to supply a BP neural network with sufficient training samples, two sample expansion techniques named the Bootstrap technique and the kernel-based Latin hypercube sampling technique are proposed and compared.Based on these proposed sample-expanding techniques,two different groups of data sample can be obtained for neural networks training.Then,the trained neural networks are utilized for the reliability prediction and the proper sample expanding technique is then selected according to the prediction accuracy.Finally,an example is included for simulation.It is shown that the method combing the BP neural network with the sample-expanding technique has more accurate prediction.
In system planning, traditional reliability evaluation methods are mainly based on logical reasoning or statistical analytics which needs enough ‘fault-outage' observations data samples for modeling. Therefore, reliability indices cannot be accurately estimated with dynamic changes of distribution system structure and insufficient data records. This paper proposed a BP-neural-network-based reliability forecasting method by considering the statistics of small sample. Firstly, in order to supply a BP neural network with sufficient training samples, two sample expansion techniques named the Bootstrap technique and the kernel-based Latin hypercube sampling technique are proposed and compared.Based on these proposed sample-expanding techniques,two different groups of data sample can be obtained for neural networks training.Then,the trained neural networks are utilized for the reliability prediction and the proper sample expanding technique is then selected according to the prediction accuracy.Finally,an example is included for simulation.It is shown that the method combing the BP neural network with the sample-expanding technique has more accurate prediction.
引文
[1]孙腾飞,程浩忠,张立波,等.基于改进混合抽样与最小切负荷计算的电力系统可靠性评估方法[J].电力系统保护与控制,2016,44(13):96-103.SUN Teng-Dei,CHENG Hao-Zhong,ZHANG Li-Bo,et al.Reliability evaluation based on modified mixed sampling and minimum load-cutting method[J].Power System Protection and Control,2016,44(13):96-103.
[2]Huang JH,Ge S Y,Han J,et al.A diagnostic method for distribution networks based on power supply safety standards[J].Protection and Control of Modern Power Systems,2016,1(1):1-9.
[3]赵洪山,赵航宇.考虑元件故障率变化的配电网可靠性评估[J].电力系统保护与控制,2015,43(11):56-62.ZHAO Hong-Shan,ZHAO Hang-Yu.Distribution system reliability analysis considering the elements failure rate changes [J].Power System Protection and Control,2015,43(11):56-62.
[4]胡美玉,胡志坚,邓奥攀,等.基于元件层级和电源可达性的配电网可靠性评估混合算法[J].电力系统保护与控制,2016,44(8):22-29.HU Mei-Yu,HUZhi-Jian,DENG Ao-Pan,et al.Hybrid algorithm for reliability evaluation of distribution network based on element hierarchy and power accessibility[J].Power System Protection and Control,2016,44(8):22-29.
[5]曹智平,周力行,张艳萍,等.基于供电可靠性的微电网规划[J].电力系统保护与控制,2015,43(14):10-15.CAO Zhi-Ping,ZHOU Li-Xing,ZHANG Yan-Ping,et al.Micro-grid planning based on supply reliability[J].Power System Protection and Control,2015,43(14):10-15.
[6]熊小伏,王建,袁峻,等.时空环境相依的电网故障模型及在电网可靠性评估中的应用[J].电力系统保护与控制,2015,43(15):28-35.XIONG Xiao-Fu,WANG Jian,YUAN Jun,et al.Temporal and spatial environments dependent power grid failure method and its application in power grid reliability assessment[J].Power System Protection and Control,2015,43(15):28-35.
[7]Billington R,Allan R N.Reliability evaluation of power systems[M].New York And London:Plenum Press,1996.
[8]袁文根,欧阳光,张伟,等.基于供电安全标准的配电网诊断方法[J].农村电气化,2016,4(5):16-18.YUAN Wen-gen,OU Yang-guang,ZHANG Wei,et al.A diagnostic method for distribution networks based on power supply safety standards[J].Rural Electrification,2016,4(5):16-18.
[9]Moradkhani A,Haghifam M R,Mohammadzadeh M.Bayesian estimation of overhead lines failure rate in electrical distribution systems[J].International Journal of Electrical Power & Energy Systems,2014,56:220-227.
[10]Xie Kai-Gui,hang H,Singh C.Reliability forecasting models for electrical distribution systems considering component failures and planned outages[J].International Journal of Electrical Power & Energy Systems,2016,79:228-234.
[11]Efron B.Bootstrap methods:another look at the jackknife[J].Annals of Statistics,1979,7(1):1-26.
[12]Hagan M T,Menhaj M B.Training feedforward networks with the Marquardt algorithm[J].IEEE Transactions on Neural Networks,1994,5(6):989-993.
[13]Billinton R,Huang D.Aleatory and epistemic uncertainty considerations in power system reliability evaluation[C].International Conference on Probabilistic Methods Applied to Power System,Rincon,Puerto Rico,2008.
[14]You Da-hai,Chen Qing-qian,Yin Xiang-gen,et al.A study of electrical security risk assessment system based on electricity regulation[J].Energy Policy,2011,39(4):2 062-2 074.
[15]田洪迅,王宏刚,万涛,等.基于BP神经网络的配电网可靠性关联因素灵敏度计算方法[J].电力系统保护与控制,2017,44(19):71-77.TIAN Hong-xun,WANG Hong-gang,WAN Tao,et al.A methodology of computing sensitivity of distribution system reliability with respect to driving factors based on a BP neural network[J].Power System Protection and Control,2017,44(19):71-77.
[2]Huang JH,Ge S Y,Han J,et al.A diagnostic method for distribution networks based on power supply safety standards[J].Protection and Control of Modern Power Systems,2016,1(1):1-9.
[3]赵洪山,赵航宇.考虑元件故障率变化的配电网可靠性评估[J].电力系统保护与控制,2015,43(11):56-62.ZHAO Hong-Shan,ZHAO Hang-Yu.Distribution system reliability analysis considering the elements failure rate changes [J].Power System Protection and Control,2015,43(11):56-62.
[4]胡美玉,胡志坚,邓奥攀,等.基于元件层级和电源可达性的配电网可靠性评估混合算法[J].电力系统保护与控制,2016,44(8):22-29.HU Mei-Yu,HUZhi-Jian,DENG Ao-Pan,et al.Hybrid algorithm for reliability evaluation of distribution network based on element hierarchy and power accessibility[J].Power System Protection and Control,2016,44(8):22-29.
[5]曹智平,周力行,张艳萍,等.基于供电可靠性的微电网规划[J].电力系统保护与控制,2015,43(14):10-15.CAO Zhi-Ping,ZHOU Li-Xing,ZHANG Yan-Ping,et al.Micro-grid planning based on supply reliability[J].Power System Protection and Control,2015,43(14):10-15.
[6]熊小伏,王建,袁峻,等.时空环境相依的电网故障模型及在电网可靠性评估中的应用[J].电力系统保护与控制,2015,43(15):28-35.XIONG Xiao-Fu,WANG Jian,YUAN Jun,et al.Temporal and spatial environments dependent power grid failure method and its application in power grid reliability assessment[J].Power System Protection and Control,2015,43(15):28-35.
[7]Billington R,Allan R N.Reliability evaluation of power systems[M].New York And London:Plenum Press,1996.
[8]袁文根,欧阳光,张伟,等.基于供电安全标准的配电网诊断方法[J].农村电气化,2016,4(5):16-18.YUAN Wen-gen,OU Yang-guang,ZHANG Wei,et al.A diagnostic method for distribution networks based on power supply safety standards[J].Rural Electrification,2016,4(5):16-18.
[9]Moradkhani A,Haghifam M R,Mohammadzadeh M.Bayesian estimation of overhead lines failure rate in electrical distribution systems[J].International Journal of Electrical Power & Energy Systems,2014,56:220-227.
[10]Xie Kai-Gui,hang H,Singh C.Reliability forecasting models for electrical distribution systems considering component failures and planned outages[J].International Journal of Electrical Power & Energy Systems,2016,79:228-234.
[11]Efron B.Bootstrap methods:another look at the jackknife[J].Annals of Statistics,1979,7(1):1-26.
[12]Hagan M T,Menhaj M B.Training feedforward networks with the Marquardt algorithm[J].IEEE Transactions on Neural Networks,1994,5(6):989-993.
[13]Billinton R,Huang D.Aleatory and epistemic uncertainty considerations in power system reliability evaluation[C].International Conference on Probabilistic Methods Applied to Power System,Rincon,Puerto Rico,2008.
[14]You Da-hai,Chen Qing-qian,Yin Xiang-gen,et al.A study of electrical security risk assessment system based on electricity regulation[J].Energy Policy,2011,39(4):2 062-2 074.
[15]田洪迅,王宏刚,万涛,等.基于BP神经网络的配电网可靠性关联因素灵敏度计算方法[J].电力系统保护与控制,2017,44(19):71-77.TIAN Hong-xun,WANG Hong-gang,WAN Tao,et al.A methodology of computing sensitivity of distribution system reliability with respect to driving factors based on a BP neural network[J].Power System Protection and Control,2017,44(19):71-77.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |