铁道科学与工程学报2024，Vol.21Issue(7)：2871-2883,13.DOI:10.19713/j.cnki.43-1423/u.T20231656
严重遮挡场景下AOA-ENN辅助列车定位的方法研究
AOA-ENN assisted train positioning in severe occlusion scenarios
摘要
Abstract
The satellite occlusion situation around the railway is complicated and variable.When trains operate in severe occlusion scenarios such as tunnels,the Beidou satellite navigation system/strapdown inertial navigation system(BDS/SINS)integrated train positioning system cannot receive satellite signals,resulting in the accumulation of train positioning errors and even positioning failure.To improve the positioning accuracy of the train in severe occlusion scenarios,an Elman neural network optimized by Archimedes optimization algorithm(AOA-ENN)was proposed to assist the BDS/SINS integrated train positioning system.Firstly,the innovation theory was introduced into the unscented Kalman filter algorithm to obtain the adaptive unscented Kalman filter algorithm(AUKF),which was used as the information fusion algorithm of the BDS/SINS integrated train positioning system.Secondly,the train operation scenarios recognition model was established based on the fuzzy C-means clustering algorithm(FCM),which could recognize train operation scenarios autonomously according to the environmental characteristic parameters.Finally,the scenarios recognition model outputted judgment results.The train run in the open scenarios,low-occlusion scenarios,and high-occlusion scenarios,the positioning information of BDS.The SINS was fused by AUKF to complete the train positioning.At the same time,the collected train positioning data was added to the training set for online training of AOA-ENN.When the train was running in severe occlusion scenarios,BDS could not receive the signal normally.The trained AOA-ENN was used to assist the integrated train positioning system.The AUKF was used to fuse the prediction information of AOA-ENN and the information calculated by SINS to gain the positioning results.The experimental results indicate that the positioning success rate of the AOA-ENN assisted train integrated positioning system reaches 98.2%in severe occlusion scenarios.Simultaneously,through the comparison experiments of different optimization algorithms and neural networks,the superiority of AOA-ENN in assisting the positioning of the integrated train positioning system is verified.The results can provide a reference for optimizing the positioning accuracy of trains in severe occlusion scenarios such as tunnels.关键词
列车组合定位系统/运行环境识别/自适应无迹卡尔曼滤波/阿基米德优化算法/Elman神经网络Key words
integrated train positioning system/operation scenarios recognition/adaptive unscented Kalman filter/Archimedes optimization algorithm/Elman neural network分类
交通工程引用本文复制引用
武晓春,杨伟康..严重遮挡场景下AOA-ENN辅助列车定位的方法研究[J].铁道科学与工程学报,2024,21(7):2871-2883,13.基金项目
中国国家铁路集团有限公司基金资助项目(N2022G012) （N2022G012）
国家自然科学基金资助项目(61661027) （61661027）
