容性电气设备绝缘性能在线检测方法的研究

发布时间：2019-04-16 14:33

【摘要】：绝缘材料作为电气设备重要的组成部分,承担着隔离不同电位,保证设备正常运转的作用,在很大程度上决定着电气设备的使用寿命,而在高压变电站中电容型设备的数量通常要占到整个变电站电气设备总数的40%～50%,确保电容型设备的绝缘良好对保证电力系统的安全可靠运行具有重要作用。目前绝大多数的电容型设备绝缘性能在线检测方法都是以绝缘介质损耗因数作为首选测量参量,因介损值较小,且易受电网干扰因素的影响,因此寻找高精度的测量方法仍具有重要的意义。本文采用的基于Hilbert变换的绝缘介质损耗因数的计算方法不受电网频率波动和非整周期采样的影响,仿真最大绝对误差处于10-5的数量级,考虑到电网谐波对计算结果的影响,在电压总谐波畸变率处于正常范围内时,这种计算方法依然能够保证很好的计算精度,误差水平几乎不变。为获取用于介损值计算的采样电压和电流信号中的基波分量以及处理介损值的在线监测数据,我们采用基于非线性自回归神经网络(NAR)数据预测端点延拓的经验模态分解法,这种信号分解的方法尤其适合处理非线性、非平稳的数据序列,其效果要好于经常被采用的对称极值方法,在电网波形受到较大扰动而波动严重的情况下,依然能准确地分解出基波分量。同时为进一步减小测量误差,消除噪声的影响,在采样数据的预处理阶段,我们采用数学形态学交替混合滤波器和小波降噪的滤波方案。采用这些方法逐步处理采样数据,计算出介质损耗因数,仿真试验表明,误差处于介损值测量要求的范围内,能够达到预期的精度。
[Abstract]:As an important part of electrical equipment, insulating material plays an important role in isolating different potential and ensuring the normal operation of the equipment, which determines the service life of electrical equipment to a great extent. In high voltage substations, the number of capacitive equipment usually accounts for 40% / 50% of the total electrical equipment in the substation. It is very important to ensure the good insulation of capacitive equipment to ensure the safe and reliable operation of power system. At present, most of the on-line testing methods for insulation performance of capacitive equipment are based on dielectric loss factor as the first choice measurement parameter, because the dielectric loss value is small, and is easily affected by the interference factors of the power network. Therefore, it is still of great significance to search for high-precision measurement methods. The calculation method of dielectric loss factor based on Hilbert transform in this paper is not affected by the fluctuation of power network frequency and non-integral period sampling. The maximum absolute error of simulation is in the order of 10 ~ 5, considering the influence of power network harmonic on the calculation results. When the total voltage harmonic distortion rate is in the normal range, the calculation method can still ensure a good calculation accuracy, and the error level is almost unchanged. In order to obtain the fundamental components of sampled voltage and current signals for dielectric loss calculation and the on-line monitoring data for processing dielectric loss values, we adopt the empirical mode decomposition method based on nonlinear autoregressive neural network (NAR) data to predict endpoint continuation. The method of signal decomposition is especially suitable to deal with nonlinear and non-stationary data series, and its effect is better than that of symmetric extremum method which is often used. When the power network waveform is disturbed and fluctuated seriously, the result is better than that of the symmetric extremum method which is often used. The fundamental component can still be decomposed accurately. At the same time, in order to further reduce the measurement error and eliminate the influence of noise, in the pre-processing stage of sampled data, we adopt the alternating hybrid filter of mathematical morphology and the filtering scheme of wavelet de-noising. These methods are used to process the sampled data step by step and calculate the dielectric loss factor. The simulation results show that the error is within the required range of dielectric loss measurement and can achieve the desired accuracy.
【学位授予单位】：中国石油大学(华东)
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM934.3

【相似文献】