面向城轨交通大数据的信息融合处理与感知技术OA

In allusion to the problem of massive multi-source heterogeneous data processing in the intelligent operation and maintenance process of urban rail transit,an urban rail transit information processing system framework based on multi-level information fusion and fine state perception is proposed.In terms of data processing,a method of data fusion based on improved D-S evidence theory is designed,and the Jousselme distance is used to optimize the conflict measurement and evidence weight distribution,which effectively improves the reliability of multi-source data.After data preprocessing,a long short term memory(LSTM)network model combined with attention mechanism is constructed,which can adaptively extract key state features from time-series data.For the feature classification,a hybrid state perception model based on fuzzy C-means clustering and support vector machine(SVM)is proposed.By fusing fuzzy membership information and accurate classification mechanism,the accurate evaluation and early warning of system health state are realized.The experimental results on public and self-built datasets show that the proposed technical solution can realize a state recognition accuracy of 93.5%,which is about 1.0%higher than the suboptimal CNN-LSTM comparison method.The average fault warning time is 35.2 min,which is 11.0%and 56.4%higher than the traditional CNN-LSTM and SVM models,respectively.It provides a new technical approach for improving the intelligent operation and maintenance level of urban rail transit systems.