摘要
针对高铁列车运行时,受电弓与接触网接触不良而产生燃弧现象,实现燃弧检测高精度、高效率的问题,该文设计了一种基于MobileNet-SSD的燃弧检测模型。通过对受电弓数据集图片进行旋转、剪切、加噪等处理,提高模型的鲁棒性;针对SSD网络参数计算量大、模型复杂的问题,去掉MobileNet网络的池化层、全连接层和Softmax层,与SSD网络的多尺度特征映射层相连接。加入感受野增大模块,提高模型提取特征信息的能力,并通过BN算法对数据信息做归一化处理,降低样本间的差异性。将受电弓燃弧数据集输入改进后的模型做对比实验,结果显示,改进SSD算法检测速度最快可以达到46.19帧/s,均值平均精度相比原算法提升7.73%,该模型可以满足列车弓网燃弧检测需求,具有较高的应用价值。
To address the problem of achieving high accuracy and efficiency in arc ignition detection due to poor contact between pantographs and contact networks during the operation of high-speed trains,the authors of this paper design a MobileNet-SSD based arc ignition detection model.By rotating,cropping and adding noise to the pantograph data set images,the robustness of the model is improved.To address the problems of large parameter calculation and complex model of SSD network,the pooling layer,fully connected layer and Softmax layer of MobileNet network are removed and connected with the multi-scale feature mapping layer of SSD network.The perceptual field increase module is added to improve the model's ability to extract feature information,and the BN algorithm is used to normalize the data information and reduce the variability between samples.The pantograph arc-firing dataset was fed into the improved model for comparative experiments.The results show that the improved SSD algorithm can detect as fast as 46.19 frames/s,and the mean average accuracy is improved by 7.73%compared with the original algorithm,which means that the model can meet the demand for arc burning detection in train arch networks and has high application value.
作者
冯庆胜
董琦
刘雨奇
付明雨
FENG Qingsheng;DONG Qi;LIU Yuqi;FU Mingyu(School of Automation and Electrical Engineering,Dalian Jiaotong University,Dalian 116028,China;School of Computer and Communication Engineering,Dalian Jiaotong University,Dalian 116028,China)
出处
《自动化与仪表》
2023年第11期54-59,共6页
Automation & Instrumentation
