基于双目立体视觉的变电站空间毫米级区域故障三维场域监测方法
    点此下载全文
引用本文:李文伟1,朱龙2,陈子文2.基于双目立体视觉的变电站空间毫米级区域故障三维场域监测方法[J].计算技术与自动化,2023,(4):99-104
摘要点击次数: 83
全文下载次数: 0
作者单位
李文伟1,朱龙2,陈子文2 (1.广西电网有限责任公司钦州供电局广西 钦州 5350002.南京南瑞继保电气有限公司江苏 南京 530023) 
中文摘要:针对变电站三类基础故障复杂度较高、无法定位空间毫米级区域故障、故障分析不全面、检测精度低等问题,提出了基于双目立体视觉的变电站空间毫米级区域故障三维场域监测方法。该方法采用双目立体视觉建立成像模型,对该模型校订后进行变电站空间毫米级区域故障三维场域图像的采集与处理,以此消除噪声污染,提升监测效果;依据结果基于改进Hu不变矩提取故障特征向量,让其在正则化极限学习机模型中训练并得出最终结果,从而实现该监测。实验结果表明,通过对该方法开展对比测试,验证该方法的图像质量最高,对六种故障状态的检测准确率均维持在80%以上,最高接近100%。该方法提高了对故障监测的精度,具有较好应用前景。
中文关键词:双目立体视觉  变电站空间毫米级区域  故障三维场域监测  图像预处理
 
Three-dimensional Field Monitoring Method for Millimeter-level Regional Faults in Substation Space Based on Binocular Stereo Vision
Abstract:In view of the high complexity of three basic faults, inability to locate millimeter level regional faults, incomplete fault analysis and low detection accuracy, a three-dimensional field monitoring method of millimeter-level regional faults in substation space based on binocular stereo vision is proposed. This method uses binocular stereo vision to establish the imaging model, collect and processes the 3D field image of the substation space millimeter level area, to eliminate noise pollution and improve the monitoring effect. According to the results, extract the fault feature vector based on the improved Hu constant moment, train it in the regularization limit learning machine model and obtain the final result to realize the monitoring. The experimental results show that through the comparative test of the proposed method, the detection accuracy is maintained above 80%, and the highest detection accuracy is close to 100%. It improves the accuracy of fault monitoring and has a good application prospect.
keywords:binocular stereo vision  millimeter-level area of substation space  three-dimensional field monitoring of faults  image preprocessing
查看全文   查看/发表评论   下载pdf阅读器