基于改进灰色组合模型的变电设备故障率预测-高电压技术.PDF

第 43 卷 第 7 期：2249-2255 高电压技术 Vol.43, No.7: 2249-2255 2017 年 7 月 31 日 High Voltage Engineering July 31, 2017 DOI: 10.13336/j.1003-6520.hve.20170628021 基于改进灰色组合模型的变电设备故障率预测 吴广宁，倪雪松，宋臻杰，高 波 610031 （西南交通大学电气工程学院，成都 ） 摘 要：现有的变电设备故障率预测模型在预测设备故障率时，往往会出现稳定期的预测数值高于实测值而损耗 期的预测值低于实测值的现象。针对该问题，在研究变电设备累积失效规律的基础上，引入“故障分界点”和“故 2 障数据分区” 个概念，并结合灰色线性回归模型建立了一种新的变电设备故障率预测优化模型。通过数值验证 对各模型的特点和有效性进行论证。结果表明，对处于故障率稳定期和损耗期 2 个阶段的案例，故障分界点的确 定和故障数据分区均有利于提高变电设备故障率预测的精度，改进模型的相对误差率较灰色线性回归模型降低了 3.59%，较基于M-R 算法的故障预测模型降低了 3.91%，整体拟合效果也更理想。 关键词：灰色线性回归模型；累积失效；初始数据预处理；特征点检测；故障分界点；故障数据分区；故障率预测 Prediction for Substation Equipment Failure Rate Based on Improved Grey Combination Model WU Guangning, NI Xuesong, SONG Zhenjie, GAO Bo (School of Electrical Engineering College, Southwest Jiaotong University, Chengdu 610031, China) Abstract ：When the existing failure prediction models of substation equipment are used to predict failure rate, such re- sults often appears that the fault stabilization period data are higher than the actual value and the fault loss period data are lower. For this phenomenon, on the basis of the research of development process of substation equipment failure, we in- troduced two concepts of “fault demarcation point” and “failure data partition”. Moreover, by combination with gray-linear regression model, we established a new optimization model of substation equipment failure rate prediction. Through numerical validation to the discussion of various characteristics and effectiveness of various models, the results show that fault demarcation po