矿用电连接器故障电弧信号特征及其检测方法研究

Electrical connector series arc fault is a common fault in coal mine power supply system. It is easy to cause the outage and even fire and other safety accidents. At present, there is still no effective solution to the problem of the detection of series arc fault in mine. The research method of combining experimental research, theoretical analysis and field test is used in this project. And from the point of view of current waveform analysis of the arc fault, the recognition and detection methods of the arc fault characteristics of the mine are systematically studied. With the series arc fault experimental device which has been developed, many series arc fault experiments are carried out under different circuit and arc gap parameters, which can provide us with a lot of arc fault waveforms. And the effect of experimental parameters on arc current can be analyzed. The characteristic component of arc fault can be obtained by analyzing the series arc fault current signal with time series analysis, Hilbert-Huang transform and wavelet transform. Through the comparative analysis between field and laboratory experiments, the feasibility of signal decomposition method can be verified. By using data fusion technology, the characteristic value of arc fault can be obtained. Then the characteristic criterion and detection method of arc fault are put forward. The research results of this project will improve the detection theory of series arc fault in mine, provide theoretical basis for the manufacture of mine arc fault circuit breaker, and enhance the reliability of power supply system in coal mine.

矿用电连接器串联型故障电弧是煤矿供电系统常见的故障，易导致供电中断甚至引发火灾等安全事故，目前仍无法有效解决矿井串联型故障电弧的检测问题。项目采用实验研究、理论分析和现场测试相结合的研究方法，从分析故障电弧电流波形的角度出发，系统地研究矿井故障电弧特征的识别与检测方法。利用研制的矿井串联型故障电弧实验系统，进行不同电路参数和弧隙参数的故障电弧实验并获取大量的故障电弧波形，分析实验参数对故障电弧电流的影响；采用时间序列分析、希尔伯特黄变换、小波分解等方法从实验电流信号中分解出故障电弧特征分量；通过现场试验并与实验室结果对比分析，验证信号分解方法的可行性；采用数据融合等方法获得故障电弧的特征值；提出矿井串联型故障电弧的特征判据；研究故障电弧特征值的检测方法。项目的研究成果将完善矿井串联型故障电弧的分析与检测理论，并为研制矿井串联型故障电弧断路器提供理论依据，提高煤矿供电系统可靠性。

为减少串联型故障电弧对煤矿供电系统造成的危害，研制了3套具有不同功能的串联型故障电弧实验装置。采用实验、仿真和理论分析相结合的方法，系统的研究了串联型故障电弧的典型特征、识别方法、仿真模型、多负载回路故障电弧选线方法及单相回路串联型故障电弧定位方法。.研究发现，对于单相阻感负载，电弧稳定燃烧时，环境湿度越大，电流信号的过零点数越多，故障电弧电流峰值越小，适当增加环境湿度，有利于熄弧；发生故障电弧时，随振动频率的增大，故障电弧电阻呈减小趋势；谐波含有率、瞬时频率方差、瞬时频率频带、时频谱均值能够表征故障电弧的时域、频域特征。对于电机及变频器负载，以相邻两工频周期的电流信号的基波和各次谐波正负序分量的幅值变化率作为特征属性，利用决策树判别规则可识别串联型故障电弧；建立极限学习机故障电弧预测模型，将负载端电压小波包节点系数预测残差时频谱均值作为阈值，可有效检测故障电弧；采用基于电流时间序列数据相空间重构以及MeanShift聚类理论，可以明显的区分出单相故障电弧、两相故障电弧和三相故障电弧。对于单相多负载回路，干路电流的关联维数、小波包-AR谱频带能量能够有效的描述多回路电路的串联型故障电弧的特征；对于三相多负载回路，采用FRFT-TBSVD的特征量提取方法提取故障电弧特征，利用经PSO-GS优化的SVM识别模型，故障电弧选线准确率超过了85%。以Mayr-Schwarz电弧数学模型为基础，建立了井下供电系统串联型故障电弧的仿真模型，对故障电弧及其他电气故障进行了仿真分析，并建立了K近邻串联型故障电弧诊断模型。此外，在获得单相回路电源端电压、电源端电流、负载端电压、负载端电流的基础上，建立了单相配电线路串联型故障电弧定位等效模型。.本项目已全面完成计划任务书所规定的研究内容，共发表学术论文35篇、授权国家发明专利5项，培养研究生17名。