基于多特征量的油中局放超声直达波识别研究

发布时间：2018-07-11 21:48

本文选题：局部放电 + 超声直达波　；参考：《中国电机工程学报》2014年03期

【摘要】：电气设备油中局部放电超声阵列定位是将阵列信号处理技术应用于局放超声检测中的一种新方法,其实质是通过相应的算法得到信号的方位信息,继而对局放源进行定位。研究表明,当超声传感器接收到直达信号时,可对局放源进行准确定位;当接收到混叠(多路径)信号时,会导致定位成功率和定位精度下降,甚至产生虚假定位。因此,有效识别超声直达波与混叠波信号是局放定位成功的关键。该文选取峰值因数、分形盒维数和李雅普指数作为局放超声直达波信号的特征参量;分别基于各特征量进行超声直达波识别的仿真研究,结果表明超声信号的混叠程度会对识别结果造成很大影响,利用BP人工神经网络,提出一种基于多特征量的电气设备油中局放超声直达波识别方法,仿真结果显示其识别成功率接近100%;最后在设定环境下进行油中局放超声直达波识别的实验研究,结果表明,采用单个特征量识别直达波和混叠波的成功率约为90%,基于多特征量的识别成功率为95%,验证了所提方法的有效性。
[Abstract]:Partial discharge ultrasonic array location in oil of electrical equipment is a new method which applies array signal processing technology to partial discharge ultrasonic detection. Its essence is to obtain the azimuth information of signal by corresponding algorithm, and then to locate the partial discharge source. The results show that when the direct signal is received by the ultrasonic sensor, the local discharge source can be accurately located, and when the mixed (multipath) signal is received, the success rate and accuracy of the location will decrease, and even the false location will be produced. Therefore, effective identification of ultrasonic direct wave and mixed wave signal is the key to the success of PD localization. In this paper, the peak factor, fractal box dimension and Leap index are selected as the characteristic parameters of PD ultrasonic direct wave signal, and the simulation research of ultrasonic direct wave recognition based on each characteristic quantity is carried out respectively. The results show that the degree of mixing of ultrasonic signals will have a great impact on the recognition results. Using BP artificial neural network, a method for identifying ultrasonic direct wave of partial discharge in oil of electrical equipment based on multiple characteristic quantities is proposed. The simulation results show that the success rate of identification is close to 100. Finally, the experimental study of direct wave recognition of partial discharge ultrasound in oil is carried out in a given environment, and the results show that, The success rate of identifying direct wave and aliasing wave by using a single feature is about 90, and the success rate of recognition based on multiple features is 95. The effectiveness of the proposed method is verified.
【作者单位】：河北省输变电设备安全防御重点实验室(华北电力大学);广西电网公司河池供电局;天津市电力公司城西供电分公司;
【基金】：国家自然科学基金项目(51307060) 国家863高技术基金项目(2011AA05A121)~~
【分类号】：TM855

【参考文献】