基于深度学习的电容器介损角在线辨识

发布时间：2018-03-17 23:31

本文选题：深度学习　切入点：电容器　出处：《电工技术学报》2017年15期 　论文类型：期刊论文

【摘要】：当前电容器介质损耗因素的计算方法为正向求解过程,即先对电容器工作电流和电压进行采样,再使用谐波分析等方法计算介损值,实践中算法稳定性不佳。为此提出了一种基于深度学习的电容器介损角辨识方法,采用一段时间的监测值训练深度学习网络,再使用该深度学习网络对新采样的信号进行辨识,判断介损角变化量(分辨率为0.001%)。给出了用于深度学习的介损角表示信号Dδ(t)的计算过程,证明了在讨论域内该信号的幅值即是介损角δ,且其波形形状包含监测装置受到的干扰。仿真实验证明该方法有效,比加汉宁窗的谐波分析法具有更好的抗噪能力。实际在线监测样本的计算结果表明其稳定性优于加汉宁窗的谐波分析法,且辨识结果不受电压互感器角差的影响。
[Abstract]:At present, the calculation method of dielectric loss factor of capacitor is the process of positive solution, that is, sampling the working current and voltage of capacitor, then calculating the dielectric loss value by harmonic analysis, etc. In practice, the stability of the algorithm is not good. In this paper, a method for identifying the dielectric loss angle of capacitors based on depth learning is proposed. The depth learning network is used to train the depth learning network with the monitoring value for a period of time, and then the new sampled signal is identified by the depth learning network. The calculation process of dielectric loss angle representation signal D 未 t for deep learning is given. It is proved that the amplitude of the signal in the discussion domain is dielectric loss angle 未, and its waveform shape includes the interference of the monitoring device. The simulation results show that the method is effective. The calculation results of on-line monitoring samples show that the stability of the method is better than that of the harmonic analysis method with Hanning window, and the identification results are not affected by the angle difference of voltage transformer.
【作者单位】：华北电力大学控制与计算机工程学院;
【基金】：国家自然科学基金(51407076,51677072) 中央高校基本科研业务专项资金(2014MS131)资助项目
【分类号】：TM53

【相似文献】