基于FastICA的盲源分离算法在相对介损监测系统中的应用

Application of BSS Algorithm Based on FastICA in On-line Monitoring System of Relative Dielectric Loss

为了提高电力系统高压电气设备监测的安全性、可靠性及信息化指标,设计了一种电容型高压设备相对介损在线监测系统。针对系统在相对介损测量过程中,由于现场噪声干扰带来的相位测量误差问题,引入独立分量分析(ICA)对电流信号进行预处理。建立了ICA算法的数学模型,给出一种基于负熵极大的FastICA算法的原理及计算步骤。对实际测量所得模拟末屏电流信号叠加高斯白噪声,仿真验证了该算法的有效性。结果表明,盲源分离算法有效地抑制了噪声,很大程度上提高了相对介损角测量的准确性。

A type of on-line monitoring system of relative dielectric loss in capacitive high-voltage apparatus is designed, with which the safety, reliability and information-based standard of monitoring high-voltage apparatus in power systems can be improved greatly. For solving the problem that field noise causes phase error to relative dielectric loss angle measurement, independent component analysis （ICA）is introduced to the pre-processing of current signals. The mathematical model of ICA algorithm is established. The principle of FastICA algorithm based on Negentropy-maximization for ICA and its calculation procedure are given. The measured signal simulating top shield current by experiment is superimposed by Gaussian white noise, and a simulation experiment verifies the effectiveness of the algorithm. The result indicates that the BSS algorithm can suppress noise effectively and improve the accuracy of relative dielectric loss angle measurement drastically.