【摘要】 该文研究基于改进XGBoost算法的XLPE电缆接头故障自动化诊断方法。以35 kV XLPE电缆接头为例，设计局放模拟实验，测量4种绝缘故障局放信号，生成二维局放图谱。从中提取描述投影形状和正负半周轮廓差异的故障特征，构建一维向量输入XGBoost模型，实现故障自动化诊断。应用哈里斯鹰算法优化模型参数，提高诊断分类性能。实验结果表明，该方法能有效测量不同故障类型的局放图谱，并以其特征实现高精度的XLPE电缆接头故障自动化诊断，确保了电缆长期稳定运行，更好地保障了电力安全。更多还原

【Abstract】 Research on an automated fault diagnosis method for XLPE cable joints based on an improved XGBoost algorithm. Taking the 35 k V XLPE cable joint as an example,design a partial discharge simulation experiment to measure four types of insulation fault partial discharge signals and generate a two-dimensional partial discharge map.Extract fault features that describe the differences in projection shape and positive and negative half circumference contours from them,construct one-dimensional vector input XGBoost model,and achieve automated fault diagnosis. Applying the Harris hawks optimization to optimize model parameters and improve diagnostic classification performance.The experimental results show that this method can effectively measure partial discharge spectra of different types of faults and achieve high-precision automatic diagnosis of XLPE cable joint faults based on their characteristics,ensuring long-term stable operation of cables and better ensuring power safety.更多还原