摘要
城市轨道交通信号系统的测试工作量大,迫切需要把一些重复的测试工作自动化,但站场图操作和显示仍需要人工操作和判定显示结果,常见自动化测试工具不能很好地支持。随着人工智能技术的发展,展现出减少信号系统测试过程中对人工依赖的前景。通过卷积神经网络(CNN)进行图像和数字信息的特征提取和识别,完成信号系统人机界面的识别和匹配,从而实现测试场景的自动执行和结果自动判断。同时对大量系统性能测试日志的特征进行提取,辅助人工发现异常特征的日志,提升性能测试效率。
The testing workload of an urban rail transit signaling system is heavy,and there is an urgent need to automate some repeated testing work.However,the operation and display of station/yard view still need manual operation and judgment of display results,which cannot be well supported by common automatic testing tools.With the development of artificial intelligence technology,the trend of reducing dependence on manpower emerges in the testing process of the signaling system.The Convolutional Neural Network(CNN)was applied to carry out feature extraction and recognition of image and digital information,complete the recognition and matching of man-machine interface of the signaling system,thus realizing the automatic execution of test scene and automatic judgment of results.The features of a great number of system performance test logs were extracted to assist the manual detection of logs with abnormal features and improve the performance test efficiency.
作者
兰青辉
刘锦峰
陈晓轩
LAN Qinghui;LIU Jinfeng;CHEN Xiaoxuan(CASCO SIGNAL LTD.,Shanghai 200071,China)
出处
《铁路技术创新》
2021年第S01期121-126,共6页
Railway Technical Innovation
关键词
轨道交通
信号系统
测试自动化
人工智能
rail transit
signaling system
test automation
artificial intelligence
