不均匀受潮对油纸绝缘频域介电特性的影响研究

Research on Effect of Non-uniform Moisture on Frequency-Domain Dielectric Characteristics of Oil-paper Insulation

摘要:

频域介电谱技术以其非破坏性及多参数分析特性，在油纸绝缘受潮评估领域具有重要应用价值。现有研究多基于水分均匀分布的假设进行整体含水量估算，忽略了实际工况中水分不均匀分布的影响。针对水分非均匀分布对介电响应的影响机制开展研究，通过控制受潮梯度制备了水分分布差异化的油浸纸样本，系统测试了其频域介电特性。实验结果表明，在相同总含水量条件下，复电容虚部与介质损耗因数（tan δ）的频谱特性与水分分布均匀度显著相关。随着分布不均匀性加剧，低频区介电参数值降低，而高频区则升高。进一步分析发现，低频介电响应主要受样本最干燥区域介电特性支配，高频段则敏感反映最高含水量区域的极化特性。该发现突破了传统均匀模型的理论局限，为构建基于频域介电谱的多层水分分布反演模型提供了实验依据。

Abstract:

Frequency domain dielectric spectroscopy (FDS) technology, with its non-destructive and multi-parameter analysis characteristics, holds significant application value in moisture assessment of oil-paper insulation. Current studies mostly estimate overall moisture content under the assumption of uniform water distribution, neglecting the impact of non-uniform moisture distribution under actual operating conditions. This study investigates the influence mechanism of non-uniform moisture distribution on dielectric responses. Oil-impregnated paper samples with differentiated moisture distributions are prepared by controlling the moisture gradient, and their frequency-domain dielectric characteristics are systematically tested. Experimental results indicate that under the same total moisture content conditions, the spectral characteristics of the imaginary part of complex capacitance and dielectric loss factor (tan δ) exhibit significant correlation with moisture distribution uniformity. As the non-uniformity of distribution intensifies, dielectric parameter values decrease in the low-frequency region while increasing in the high-frequency region. Further analysis reveals that low-frequency dielectric responses are predominantly governed by the driest regions of samples, while high-frequency characteristics sensitively reflect polarization properties of the most moisture-concentrated areas. This finding transcends the theoretical limitations of traditional uniform models and provides experimental evidence for constructing a multi-layer moisture distribution inversion model based on FDS.