基于介电模量谱和分数阶模型的植物油纸绝缘状态评估

Research on Evaluation of Vegetable Oil-paper Insulation Condition Based on Dielectric Modulus Spectrum and Fractional Model

为深入研究油纸绝缘材料在宽频带内复杂的极化行为,实现植物油纸绝缘设备受潮及老化状态的量化评估,结合介电模量和分数元的概念,建立了削弱直流电导影响的分数阶模型。首先制备7组不同水分和不同老化程度的植物油纸绝缘样品,使用频域介电谱(frequency domain spectroscopy,FDS)技术测量得到频域介电谱并转换成介电模量谱;然后利用分数阶模型进行参数辨识,提出表征受潮和老化程度的特征参量,并分别与含水质量分数ωC和聚合度DP进行量化研究;最后结合1台植物油变压器验证了该评估方法的准确性。结果表明,分数阶模型能有效模拟介电模量谱的变化趋势,提出的特征参量与受潮及老化密切相关。受潮状态评估中,特征参量α、β、Ma与ωC呈幂次相关,特征参量τ与ωC呈指数相关;老化状态评估中,特征参量α、Ma、τ与DP呈幂次相关,特征参量β与DP呈指数相关。

In order to deeply study the complex behaviors of oil-paper insulation materials in a wide frequency band,and to achieve quantitative evaluation of the water content and aging state of related equipment,this paper integrates the dielectric modulus and fractional elements to establish a fractional model that weakens the influence of DC conductance.First,seven sets of vegetable oil-paper insulation samples with different moisture content and different degrees of aging were prepared,the frequency domain spectroscopy(FDS)technology was adopted to measure the frequency domain dielectric spectrum and convert it into a dielectric modulus spectrum.Then,the fractional model was adopted to identify the parameters,the characteristic quantities closely related to the water contentωC and the degree of polymerization DP were selected,and the functional relationship was established.Finally,an field vegetable oil transformer was used to measure and to preliminarily verify the effectiveness of the method.The experimental results show that the fractional model can effectively simulate the changing trend of the dielectric modulus spectrum,and the proposed characteristic parameters are closely related to moisture and aging.In the evaluation of moisture content,the characteristic parametersα,β,Ma andωC are power-related,and the characteristic parameterτis exponentially related toωC.In the evaluation of aging degree,the characteristic parametersα,Ma,τare power-related with DP,and the characteristic parameterβis exponentially related with DP.