基于鸟鸣功率谱密度的输电线路涉鸟故障相关鸟种智能识别

Intelligent Recognition of Bird Species Related to Transmission Line Bird Faults Based on Power Spectrum Density of Birdsong

鸟类活动对架空输电线路的运行影响较大,不同鸟种引起的线路故障差异明显,为了提高渉鸟故障防治的针对性,需分类识别输电线路相关鸟种。以鸟粪类、鸟巢类、鸟体短接类、鸟啄类四种渉鸟故障对应的8种代表性鸟种为例,通过分帧加窗、端点检测等预处理,采用离散傅里叶变换和Welch算法获取不同鸟鸣信号的功率谱密度(PSD)图,从中提取129个频率点的PSD特征,并基于随机森林建立分类识别模型,以PSD特征作为输入量,开展模型训练和不同鸟种分类测试,8种鸟类的识别准确率为83.3%~100.0%,提高了渉鸟故障防治措施的精准性。

Bird activities severely affect the safe operation of overhead transmission lines.Line faults caused by differ-ent bird species varies significantly.In order to improve the pertinence of bird-related outage prevention,it is necessary to conduct classification and recognition of the birds related to transmission lines.In this paper,8 kinds of representative bird species were taken for study,respectively corresponding to four types of bird-related outages.including bird excre-ment outage,nesting outage,bird short-circuit outage and pecker outage.The birdsong signals were preprocessed by framing,windowing,and endpoint detection.And then the power spectrum density(PSD)figures were obtained by dis-crete Fourier trans-formation(DEF)and Welch algorithm.The PSD features of 129 frequency points were extracted.A classification recognition model was established by random forest(RF)and the PSD features were taken as input parame-ters.The model training and classification testing of different bird species were conducted.The recog nition accuracy of 8 bird species varies from 83.3%to 100%.This study improves the pertinence of prevention countermeasures for bird relat-ed outages.