Research On Real-time Performance Optimization Of High-speed Railway Catenary Monitoring System Based On Hadoop

The good operating of the catenary is essential for the safe operation of high speed railways.It is important to monitor key parts on the catenary and to visualise the data to the user.Traditional monitoring methods are very costly and subject to weather conditions,which to some extent reduces the efficiency of catenary inspections.The real-time nature of big data operations in the catenary monitoring system determines the effectiveness of mining useful information,so real-time is an important indicator of a catenary monitoring system.This paper focuses on improving the real-time performance of big data operations in catenary monitoring systems from the perspective of artificial intelligence.The main work and results of this paper are as follows.1.A cloud-based catenary monitoring system is proposed to address the limitations of traditional catenary detection methods under the influence of weather factors.A wireless sensing protocol and open source software building blocks are used for the catenary monitoring system.Experiments show that the proposed system operates stably.2.The LSTM-RF algorithm is proposed to address the complex characteristics of catenary monitoring data.LSTM and RF are used to predict the monitoring data in stages and features,and the experiments show that the proposed algorithm improves the prediction accuracy by more than 10%.3.In response to the problem of low accuracy of the configuration parameterexecution efficiency model for big data operations,a multi-strategy-based gene expression programming algorithm is proposed.Adaptive variation rate setting and backtracking strategies are used.Experiments demonstrate that the proposed algorithm has higher model accuracy than the traditional regression algorithm in the monitoring system.4.To address the complexity of the configuration parameter-execution efficiency model for big data operations,the main swarm intelligence optimization algorithms are compared,and the algorithm suitable for the model of this topic is selected.An improved sparrow search algorithm is proposed,which avoid local optimal by using adaptive detector position update and dynamic evolutionary direction selection strategy to improve the optimization-seeking capability for the big data job modelThe algorithm proposed in this paper meets the practical engineering application of this paper,and can effectively optimise the real-time performance of big data job execution.