空气间隙的储能特征与放电电压预测

Energy Storage Features and Discharge Voltage Prediction of Air Gaps

通过数学计算而非试验研究获取空气间隙的放电电压是高电压工程领域长期以来希望达到的目标。为此,基于绝缘击穿缘于储能越限的物理思想,提出了一种空气间隙放电电压预测的新思路:将复杂的空气放电过程研究前移至间隙结构储能状态及其影响因素研究,从电场分布、冲击电压波形两个方面定义了表征空气间隙储能状态的特征集,采用支持向量机(SVM)建立了空气间隙的放电电压预测模型。利用该模型,成功实现了不同间隙长度的棒-板、棒-棒长空气间隙在不同波前时间的正、负极性操作冲击电压作用下的50%放电电压预测,4组测试样本预测结果的平均绝对百分比误差分别为3.6%、3.25%、3.5%和3.8%。该方法有助于推进电气外绝缘的数字化设计,可为构建"计算高电压工程"学科体系提供参考。

It has been a long sought goal for the field of high voltage engineering to obtain the discharge voltage of air gaps by mathematical calculations instead of experiments. Therefore, a new idea for the discharge voltage prediction of air gaps was proposed in this paper, which is based on the physical thought that the insulation breakdown is due to the out-of-limit of the stored energy. The complicated air discharge process study was moved forward to the study of the energy storage status of the gap structure and its influence factors. The feature set used to characterize the energy storage status of air gap was defined from two aspects, including the electric field distribution and the impulse voltage waveform. The air gap discharge voltage prediction model was established by support vector machine （SVM）. By the proposed model, the 50% discharge voltage prediction of rod-plane and rod-rod long air gaps with different gap lengths was successfully achieved, under positive and negative switching impulse voltage with different wavefronts. The mean absolute percentage errors of the predicted results of 4 test sample sets are respectively 3.6%, 3.25%, 3.5% and 3.8%. This method contributes to promoting the digital design of external insulation, and provide reference for establishing the discipline system of computational high voltage engineering.