摘要
接触网常年受自然环境的影响,导致其部件容易出现异常,威胁高速列车运行安全。目前,接触网的故障检测主要采用人工现场巡检和查看图像的方式,劳动强度大、人工成本高。利用人工智能技术实现高效率的接触网故障检测,具有重要现实意义和经济价值。本文主要基于高铁线路4C装置安全巡检图像,提出一种接触网紧固件异常检测方法,首先利用基于注意力机制改进的Faster R-CNN算法,准确地实现各种紧固件的识别与定位,然后利用全卷积孪生网络实现接触网异常检测。实验表明,该方法对5种紧固件的4C异常检测精度较高。
The catenary is affected by the natural environment all the year round,which leads to the abnormality of its components and threatens the safety of high-speed trains.At present,the fault detection of catenary is mainly carried out by manual on-site inspection and image viewing,which has high labor intensity and high labor cost.It is of great practical significance and economic value to use artificial intelligence technology to realize high-efficiency fault detection of catenary.Based on the safety inspection image of 4C device of high-speed railway line,this paper proposes a detection method for abnormal fastener of catenary.Firstly,Faster R-CNN algorithm based on attention mechanism is used to accurately identify and locate various fasteners,and then full convolution twin network is applied to detect catenary defects.The experimental results show that the proposed method has high accuracy for 4C anomaly detection of five kinds of fasteners.
出处
《电气化铁道》
2020年第S02期220-225,228,共7页
Electric Railway
