摘要:
因接触不良或机械损伤，煤矿供电系统会产生串联型故障电弧，串联型故障电弧是引发煤矿电气火灾的原因之一，目前缺少有效的检测手段，影响了井下的供电安全。井下可能存在瓦斯和煤尘等易燃易爆物质，不宜开展串联型故障电弧实验，无法获得大量故障电弧电流样本，加大了井下串联型故障电弧诊断工作的难度。为研究矿井供电系统串联型故障电弧的典型特征及诊断方法，论文首先在Mayr-Schwarz电弧数学模型的基础上，建立了矿井供电系统串联型故障电弧仿真模型，并结合实验结果对仿真模型的性能进行了评估;然后对矿井供电系统的采煤系统、胶带输送系统、泵房排水系统、照明系统的串联型故障电弧、过电压、单相接地、两相接地短路、两相短路、三相短路电气故障进行仿真分析、特征分析，以电流信号的过零点数、峰峰值、方差、峭度系数、裕度因子、谐波畸变率、单边功率谱频率方差、小波包系数能量熵、小波包系数峰峰值为特征量，建立了矿井供电系统串联型故障电弧特征参数数据库;最后综合比较决策树、K近邻、Bagged trees多分类模式识别方法在故障电弧诊断、选相及抗负载电流波动扰动、抗背景噪声扰动方面的性能，提出了K近邻矿井供电系统串联型故障电弧诊断方法。结果表明，建立的串联型故障电弧仿真模型能够用于仿真分析矿井供电系统串联型故障电弧，所建立的特征参数数据库能够反映矿井供电系统串联型故障电弧的典型特征，提出的K近邻串联型故障电弧诊断方法可用于矿井供电系统串联型故障电弧诊断及选相。
Abstract:
Because of bad contact or mechanical damage,the coal power supply system might produce series arc fault. Series arc fault is one of the causes of electrical fire in coal mines. There may be inflammable and explosive materials such as gas and coal dust in the underground. It is not suitable to carry out series arc fault experiment,and thus a lot of fault diagnosis samples cannot be obtained,which brings difficulties for arc fault diagnosis. In order to study the typ- ical characteristics and diagnostic methods of series fault arcs in mine power supply system,a series arc fault simula- tion model based on the Mayr-Schwarz arc mathematical model was established for mine power supply system and the performance of this simulation model was evaluated in this paper. Then it simulated and analyzed the electrical faults such as series arc fault,over voltage,single phase ground,two phase ground short circuit,two phase short circuit,three phase short circuit in the coal mining system,belt conveyor system,pump house drainage system,lighting system of mine power supply system. Also,the series arc fault characteristic parameter database of mine power supply system was established by using Zero-crossing point,peak-to-peak value,variance,kurtosis coefficient,margin factor,harmonic dis- tortion rate,unilateral power spectrum frequency variance,wavelet packet coefficient energy entropy,wavelet packet coefficient peak-to-peak value as feature variables. Finally,the paper comprehensively compared the performance of decision tree,K-nearest neighbor and Bagged trees multi-classification pattern recognition method in fault arc diagno- sis,phase selection and anti-load current fluctuation disturbance,anti-background noise disturbance,and proposed se- ries arc fault diagnosis method for K nearest neighbor mine power supply system. The results show that the established series arc fault simulation model can be used to simulate and analyze the series arc fault of mine power supply system and the database established in this paper can reflect the typical characteristics of series arc fault in mine power supply system. It also showed that K-nearest series arc fault diagnosis method can be used for the series arc fault diagnosis and phase selection of mine power supply system.
百度学术
下载:
