基于小波与栈式稀疏自编码器的电力电缆早期故障定位方法研究
|
摘要
随着电力电缆的应用日益广泛,电力电缆早期故障的发生日趋频繁,对早期故障定位精度以及速度的要求日益提高。为实现电力电缆早期故障的定位,结合早期故障特征,提出了一种基于小波变换方法结合栈式稀疏自编码器(SAE)的电力电缆早期故障定位方法。该方法利用小波变换对故障信号进行特征提取,并将提取到的特征作为SAE的输入,依靠SAE网络强大的学习和预测能力,实现故障特征信号到故障距离的一一对应,解决电力电缆早期故障定位问题。仿真结果表明,利用小波变换提取到的故障特征训练后的神经网络可以快速准确的实现对电力电缆早期故障的定位。
With the increasing application of power cable,incipient failure of power cable is becoming more frequent,the requirements for the accuracy and speed of incipient fault location are also increasing.In order to achieve the location of incipient fault for power cable,combined with the features of incipient failures,this paper proposes an incipient fault location method for power cable based on wavelet transform,combined with stacked autoencoder.This method uses wavelet transform to extract low-level features of fault signals,and uses the extracted features as input of the stacked autoencoder.By using the powerful learning and predictive capabilities of neural networks,the relationship between the fault signal and the fault distance has been found,and the problem of incipient fault location of the power cable has been solved.Simulation results showed that the neural network,trained by fault features extracted by wavelet transform,can quickly and accurately locates the incipient faults of power cable.
With the increasing application of power cable,incipient failure of power cable is becoming more frequent,the requirements for the accuracy and speed of incipient fault location are also increasing.In order to achieve the location of incipient fault for power cable,combined with the features of incipient failures,this paper proposes an incipient fault location method for power cable based on wavelet transform,combined with stacked autoencoder.This method uses wavelet transform to extract low-level features of fault signals,and uses the extracted features as input of the stacked autoencoder.By using the powerful learning and predictive capabilities of neural networks,the relationship between the fault signal and the fault distance has been found,and the problem of incipient fault location of the power cable has been solved.Simulation results showed that the neural network,trained by fault features extracted by wavelet transform,can quickly and accurately locates the incipient faults of power cable.
引文
[1]姚海燕,张静,留毅,等.基于多尺度小波判据和时频特征关联的电缆早期故障检测和识别方法[J].电力系统保护与控制,2015,43(9):115-123.
[2] MOGHE R,MOUSAVI M J,STOUPIS J,et al.Field investigation and analysis of incipient faults leading to a catastrophic failure in an underground distribution feeder[C].IEEE/PES Power SYStems Conference and Exposition.Seattle,WA,USA:IEEE,2009:1-6.
[3]汪梅,曲立娜.基于小波和自组织网络的电缆故障识别[J].振动、测试与诊断,2009,29(3):313-316.
[4] STRINGER N T,KOJOVIC L A.Prevention of underground cable splice failures[J].IEEE Transactions on Industry Applications,2001,37(1):230-239.
[5]赵军玉,杨淑英.树型分支配电网的行波故障定位方法研究[J].智慧电力,2012,40(5):39-42.
[6]周娟娟,李虹.改进双曲S变换与动态测度结合在电能质量检测中的应用[J].智慧电力,2015,43(1):33-38.
[7]戴铭.10kV地下电缆早期故障检测与识别方法探讨[D].成都:西南交通大学,2012.
[8]庞正.基于小波分析的电缆故障测距方法研究[D].锦州:辽宁工业大学,2016.
[9]刁彦华,王玉田,陈国通.基于小波变换模极大值的信号奇异性检测[J].河北工业科技,2004,21(1):1-3.
[10]蔡维,张雪松,刘承志,等.基于PT在线注入信号及神经网络跟踪小波变换模极大值变化的配电网故障定位方法[J].电力系统保护与控制,2006,34(23):44-48.
[11]刘嫣,汪梅,杨存军.一种新的电缆单端故障测距方法研究[J].仪器仪表学报,2006,27(s1):44-45.
[12]臧川,江冰,薛心怡,等.基于小波优化神经网络的故障定位算法研究[J].电子技术应用,,2014,40(06):55-58.
[13]沈桂鹏,杨店飞,郭宇杰.基于自适应权重粒子群优化BP神经网络的光伏阵列故障诊断与定位[J].陕西电力,2016,44(8):23-27.
[14]何巨龙,王根平,刘丹,等.基于提升小波和改进BP神经网络的配电网系统电能质量扰动定位与识别[J].电力系统保护与控制,2017,45(10):69-76.
[15]陈仁祥,杨星,杨黎霞,等.栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断[J].振动与冲击,2017,36(21):125-131.
[16]徐德荣,陈秀宏,田进.稀疏自编码和Softmax回归的快速高效特征学习[J].传感器与微系统,2017,36(5):55-58.
[17]马子骥,钟广超,刘宏立,等.小波变换的稀疏最优化信号趋势项提取方法[J].传感器与微系统,2017,36(1):27-30.
[18]廖晓辉,梁恒娜,丁倩.基于小波变换的电力电缆故障测距研究[J].郑州大学学报(工学版),2013,34(3):6-9.
[19] EKICI S,YLDRM S,POYRAZ M.A neural network based approach for transmission line faults[C].5th International Conference on Electrical and Electronics Engineering,2007:229-233.
[20]王勇,赵俭辉,章登义,等.基于稀疏自编码深度神经网络的林火图像分类[J].计算机工程与应用,2014,50(24):173-177.
[21]林少飞,盛惠兴,李庆武.基于堆叠稀疏自动编码器的手写数字分类[J].微处理机,2015(1):47-51.
[22]孙文珺,邵思羽,严如强.基于稀疏自动编码深度神经网络的感应电动机故障诊断[J].机械工程学报,2016,52(9):65-71.
[23]范李平,袁兆强,张凯.基于小波变换的单相接地故障电弧模型及其PSCAD/EMTDC仿真研究[J].电力系统保护与控制,2011,39(5):51-56.
[2] MOGHE R,MOUSAVI M J,STOUPIS J,et al.Field investigation and analysis of incipient faults leading to a catastrophic failure in an underground distribution feeder[C].IEEE/PES Power SYStems Conference and Exposition.Seattle,WA,USA:IEEE,2009:1-6.
[3]汪梅,曲立娜.基于小波和自组织网络的电缆故障识别[J].振动、测试与诊断,2009,29(3):313-316.
[4] STRINGER N T,KOJOVIC L A.Prevention of underground cable splice failures[J].IEEE Transactions on Industry Applications,2001,37(1):230-239.
[5]赵军玉,杨淑英.树型分支配电网的行波故障定位方法研究[J].智慧电力,2012,40(5):39-42.
[6]周娟娟,李虹.改进双曲S变换与动态测度结合在电能质量检测中的应用[J].智慧电力,2015,43(1):33-38.
[7]戴铭.10kV地下电缆早期故障检测与识别方法探讨[D].成都:西南交通大学,2012.
[8]庞正.基于小波分析的电缆故障测距方法研究[D].锦州:辽宁工业大学,2016.
[9]刁彦华,王玉田,陈国通.基于小波变换模极大值的信号奇异性检测[J].河北工业科技,2004,21(1):1-3.
[10]蔡维,张雪松,刘承志,等.基于PT在线注入信号及神经网络跟踪小波变换模极大值变化的配电网故障定位方法[J].电力系统保护与控制,2006,34(23):44-48.
[11]刘嫣,汪梅,杨存军.一种新的电缆单端故障测距方法研究[J].仪器仪表学报,2006,27(s1):44-45.
[12]臧川,江冰,薛心怡,等.基于小波优化神经网络的故障定位算法研究[J].电子技术应用,,2014,40(06):55-58.
[13]沈桂鹏,杨店飞,郭宇杰.基于自适应权重粒子群优化BP神经网络的光伏阵列故障诊断与定位[J].陕西电力,2016,44(8):23-27.
[14]何巨龙,王根平,刘丹,等.基于提升小波和改进BP神经网络的配电网系统电能质量扰动定位与识别[J].电力系统保护与控制,2017,45(10):69-76.
[15]陈仁祥,杨星,杨黎霞,等.栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断[J].振动与冲击,2017,36(21):125-131.
[16]徐德荣,陈秀宏,田进.稀疏自编码和Softmax回归的快速高效特征学习[J].传感器与微系统,2017,36(5):55-58.
[17]马子骥,钟广超,刘宏立,等.小波变换的稀疏最优化信号趋势项提取方法[J].传感器与微系统,2017,36(1):27-30.
[18]廖晓辉,梁恒娜,丁倩.基于小波变换的电力电缆故障测距研究[J].郑州大学学报(工学版),2013,34(3):6-9.
[19] EKICI S,YLDRM S,POYRAZ M.A neural network based approach for transmission line faults[C].5th International Conference on Electrical and Electronics Engineering,2007:229-233.
[20]王勇,赵俭辉,章登义,等.基于稀疏自编码深度神经网络的林火图像分类[J].计算机工程与应用,2014,50(24):173-177.
[21]林少飞,盛惠兴,李庆武.基于堆叠稀疏自动编码器的手写数字分类[J].微处理机,2015(1):47-51.
[22]孙文珺,邵思羽,严如强.基于稀疏自动编码深度神经网络的感应电动机故障诊断[J].机械工程学报,2016,52(9):65-71.
[23]范李平,袁兆强,张凯.基于小波变换的单相接地故障电弧模型及其PSCAD/EMTDC仿真研究[J].电力系统保护与控制,2011,39(5):51-56.